flake: add platformdirs artiq dependency

Causes linker errors for unknown reasons.

.clang-format

@@ -0,0 +1,32 @@
+BasedOnStyle: LLVM
+
+Language: Cpp
+Standard: Cpp11
+
+AccessModifierOffset: -1
+AlignEscapedNewlines: Left
+AlwaysBreakAfterReturnType: None
+AlwaysBreakTemplateDeclarations: Yes
+AllowAllParametersOfDeclarationOnNextLine: false
+AllowShortFunctionsOnASingleLine: Inline
+BinPackParameters: false
+BreakBeforeBinaryOperators: NonAssignment
+BreakBeforeTernaryOperators: true
+BreakConstructorInitializers: AfterColon
+BreakInheritanceList: AfterColon
+ColumnLimit: 120
+ConstructorInitializerAllOnOneLineOrOnePerLine: true
+ContinuationIndentWidth: 4
+DerivePointerAlignment: false
+IndentCaseLabels: true
+IndentPPDirectives: None
+IndentWidth: 4
+MaxEmptyLinesToKeep: 1
+PointerAlignment: Left
+ReflowComments: true
+SortIncludes: false
+SortUsingDeclarations: true
+SpaceAfterTemplateKeyword: false
+SpacesBeforeTrailingComments: 2
+TabWidth: 4
+UseTab: Never

.gitignore

@@ -1,3 +1,4 @@
 __pycache__
 /target
+/nac3standalone/demo/linalg/target
 nix/windows/msys2

.pre-commit-config.yaml

@@ -1,24 +1,24 @@
 # See https://pre-commit.com for more information
 # See https://pre-commit.com/hooks.html for more hooks
 
-default_stages: [commit]
+default_stages: [pre-commit]
 
 repos:
   - repo: local
     hooks:
       - id: nac3-cargo-fmt
         name: nac3 cargo format
-        entry: cargo
+        entry: nix
         language: system
         types: [file, rust]
         pass_filenames: false
         description: Runs cargo fmt on the codebase.
-        args: [fmt]
+        args: [develop, -c, cargo, fmt, --all]
       - id: nac3-cargo-clippy
         name: nac3 cargo clippy
-        entry: cargo
+        entry: nix
         language: system
         types: [file, rust]
         pass_filenames: false
         description: Runs cargo clippy on the codebase.
-        args: [clippy, --tests]
+        args: [develop, -c, cargo, clippy, --tests]

Cargo.toml

@@ -4,6 +4,7 @@ members = [
   "nac3ast",
   "nac3parser",
   "nac3core",
+  "nac3core/nac3core_derive",
   "nac3standalone",
   "nac3artiq",
   "runkernel",

flake.lock

@@ -2,11 +2,11 @@
   "nodes": {
     "nixpkgs": {
       "locked": {
-        "lastModified": 1720418205,
-        "narHash": "sha256-cPJoFPXU44GlhWg4pUk9oUPqurPlCFZ11ZQPk21GTPU=",
+        "lastModified": 1731319897,
+        "narHash": "sha256-PbABj4tnbWFMfBp6OcUK5iGy1QY+/Z96ZcLpooIbuEI=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "655a58a72a6601292512670343087c2d75d859c1",
+        "rev": "dc460ec76cbff0e66e269457d7b728432263166c",
         "type": "github"
       },
       "original": {

flake.nix

@@ -6,6 +6,7 @@
   outputs = { self, nixpkgs }:
     let
       pkgs = import nixpkgs { system = "x86_64-linux"; };
+      pkgs32 = import nixpkgs { system = "i686-linux"; };
     in rec {
       packages.x86_64-linux = rec {
         llvm-nac3 = pkgs.callPackage ./nix/llvm {};
@@ -13,9 +14,24 @@
           ''
           mkdir -p $out/bin
           ln -s ${pkgs.llvmPackages_14.clang-unwrapped}/bin/clang $out/bin/clang-irrt
-          ln -s ${pkgs.llvmPackages_14.clang}/bin/clang $out/bin/clang-irrt-test
           ln -s ${pkgs.llvmPackages_14.llvm.out}/bin/llvm-as $out/bin/llvm-as-irrt
           '';
+        demo-linalg-stub = pkgs.rustPlatform.buildRustPackage {
+          name = "demo-linalg-stub";
+          src = ./nac3standalone/demo/linalg;
+          cargoLock = {
+            lockFile = ./nac3standalone/demo/linalg/Cargo.lock;
+          };
+          doCheck = false;
+        };
+        demo-linalg-stub32 = pkgs32.rustPlatform.buildRustPackage {
+          name = "demo-linalg-stub32";
+          src = ./nac3standalone/demo/linalg;
+          cargoLock = {
+            lockFile = ./nac3standalone/demo/linalg/Cargo.lock;
+          };
+          doCheck = false;
+        };
         nac3artiq = pkgs.python3Packages.toPythonModule (
           pkgs.rustPlatform.buildRustPackage rec {
             name = "nac3artiq";
@@ -24,9 +40,8 @@
             cargoLock = {
               lockFile = ./Cargo.lock;
             };
-            cargoTestFlags = [ "--features" "test" ];
             passthru.cargoLock = cargoLock;
-            nativeBuildInputs = [ pkgs.python3 pkgs.llvmPackages_14.clang llvm-tools-irrt pkgs.llvmPackages_14.llvm.out llvm-nac3 ];
+            nativeBuildInputs = [ pkgs.python3 (pkgs.wrapClangMulti pkgs.llvmPackages_14.clang) llvm-tools-irrt pkgs.llvmPackages_14.llvm.out llvm-nac3 ];
             buildInputs = [ pkgs.python3 llvm-nac3 ];
             checkInputs = [ (pkgs.python3.withPackages(ps: [ ps.numpy ps.scipy ])) ];
             checkPhase =
@@ -34,7 +49,9 @@
               echo "Checking nac3standalone demos..."
               pushd nac3standalone/demo
               patchShebangs .
-              ./check_demos.sh
+              export DEMO_LINALG_STUB=${demo-linalg-stub}/lib/liblinalg.a
+              export DEMO_LINALG_STUB32=${demo-linalg-stub32}/lib/liblinalg.a
+              ./check_demos.sh -i686
               popd
               echo "Running Cargo tests..."
               cargoCheckHook
@@ -90,18 +107,18 @@
             (pkgs.fetchFromGitHub {
               owner = "m-labs";
               repo = "sipyco";
-              rev = "939f84f9b5eef7efbf7423c735d1834783b6140e";
-              sha256 = "sha256-15Nun4EY35j+6SPZkjzZtyH/ncxLS60KuGJjFh5kSTc=";
+              rev = "094a6cd63ffa980ef63698920170e50dc9ba77fd";
+              sha256 = "sha256-PPnAyDedUQ7Og/Cby9x5OT9wMkNGTP8GS53V6N/dk4w=";
             })
             (pkgs.fetchFromGitHub {
               owner = "m-labs";
               repo = "artiq";
-              rev = "923ca3377d42c815f979983134ec549dc39d3ca0";
-              sha256 = "sha256-oJoEeNEeNFSUyh6jXG8Tzp6qHVikeHS0CzfE+mODPgw=";
+              rev = "28c9de3e251daa89a8c9fd79d5ab64a3ec03bac6";
+              sha256 = "sha256-vAvpbHc5B+1wtG8zqN7j9dQE1ON+i22v+uqA+tw6Gak=";
             })
           ];
           buildInputs = [
-            (python3-mimalloc.withPackages(ps: [ ps.numpy ps.scipy ps.jsonschema ps.lmdb nac3artiq-instrumented ]))
+            (python3-mimalloc.withPackages(ps: [ ps.numpy ps.scipy ps.jsonschema ps.lmdb ps.platformdirs nac3artiq-instrumented ]))
             pkgs.llvmPackages_14.llvm.out
           ];
           phases = [ "buildPhase" "installPhase" ];
@@ -151,7 +168,7 @@
         buildInputs = with pkgs; [
           # build dependencies
           packages.x86_64-linux.llvm-nac3
-          llvmPackages_14.clang llvmPackages_14.llvm.out  # for running nac3standalone demos
+          (pkgs.wrapClangMulti llvmPackages_14.clang) llvmPackages_14.llvm.out  # for running nac3standalone demos
           packages.x86_64-linux.llvm-tools-irrt
           cargo
           rustc
@@ -163,10 +180,12 @@
           clippy
           pre-commit
           rustfmt
-          rust-analyzer
         ];
-        # https://nixos.wiki/wiki/Rust#Shell.nix_example
-        RUST_SRC_PATH = "${pkgs.rust.packages.stable.rustPlatform.rustLibSrc}";
+        shellHook =
+          ''
+          export DEMO_LINALG_STUB=${packages.x86_64-linux.demo-linalg-stub}/lib/liblinalg.a
+          export DEMO_LINALG_STUB32=${packages.x86_64-linux.demo-linalg-stub32}/lib/liblinalg.a
+          '';
       };
       devShells.x86_64-linux.msys2 = pkgs.mkShell {
         name = "nac3-dev-shell-msys2";

nac3artiq/Cargo.toml

@@ -12,15 +12,10 @@ crate-type = ["cdylib"]
 itertools = "0.13"
 pyo3 = { version = "0.21", features = ["extension-module", "gil-refs"] }
 parking_lot = "0.12"
-tempfile = "3.10"
-nac3parser = { path = "../nac3parser" }
+tempfile = "3.13"
 nac3core = { path = "../nac3core" }
 nac3ld = { path = "../nac3ld" }
 
-[dependencies.inkwell]
-version = "0.4"
-default-features = false
-features = ["llvm14-0", "target-x86", "target-arm", "target-riscv", "no-libffi-linking"]
-
 [features]
 init-llvm-profile = []
+no-escape-analysis = ["nac3core/no-escape-analysis"]

nac3artiq/demo/list_cmp.py

@@ -0,0 +1,24 @@
+from min_artiq import *
+from numpy import int32
+
+
+@nac3
+class EmptyList:
+    core: KernelInvariant[Core]
+
+    def __init__(self):
+        self.core = Core()
+
+    @rpc
+    def get_empty(self) -> list[int32]:
+        return []
+
+    @kernel
+    def run(self):
+        a: list[int32] = self.get_empty()
+        if a != []:
+            raise ValueError
+
+
+if __name__ == "__main__":
+    EmptyList().run()

nac3artiq/demo/min_artiq.py

@@ -112,10 +112,15 @@ def extern(function):
     register_function(function)
     return function
 
-def rpc(function):
-    """Decorates a function declaration defined by the core device runtime."""
-    register_function(function)
-    return function
+
+def rpc(arg=None, flags={}):
+    """Decorates a function or method to be executed on the host interpreter."""
+    if arg is None:
+        def inner_decorator(function):
+            return rpc(function, flags)
+        return inner_decorator
+    register_function(arg)
+    return arg
 
 def kernel(function_or_method):
     """Decorates a function or method to be executed on the core device."""
@@ -201,7 +206,7 @@ class Core:
         embedding = EmbeddingMap()
 
         if allow_registration:
-            compiler.analyze(registered_functions, registered_classes)
+            compiler.analyze(registered_functions, registered_classes, set())
             allow_registration = False
 
         if hasattr(method, "__self__"):

nac3artiq/demo/str_abi.py

@@ -0,0 +1,26 @@
+from min_artiq import *
+from numpy import ndarray, zeros as np_zeros
+
+
+@nac3
+class StrFail:
+    core: KernelInvariant[Core]
+
+    def __init__(self):
+        self.core = Core()
+
+    @kernel
+    def hello(self, arg: str):
+        pass
+
+    @kernel
+    def consume_ndarray(self, arg: ndarray[str, 1]):
+        pass
+
+    def run(self):
+        self.hello("world")
+        self.consume_ndarray(np_zeros([10], dtype=str))
+
+
+if __name__ == "__main__":
+    StrFail().run()

nac3artiq/src/lib.rs

@@ -1,10 +1,4 @@
-#![deny(
-    future_incompatible,
-    let_underscore,
-    nonstandard_style,
-    rust_2024_compatibility,
-    clippy::all
-)]
+#![deny(future_incompatible, let_underscore, nonstandard_style, clippy::all)]
 #![warn(clippy::pedantic)]
 #![allow(
     unsafe_op_in_unsafe_fn,
@@ -16,63 +10,65 @@
     clippy::wildcard_imports
 )]
 
-use std::collections::{HashMap, HashSet};
-use std::fs;
-use std::io::Write;
-use std::process::Command;
-use std::rc::Rc;
-use std::sync::Arc;
+use std::{
+    collections::{HashMap, HashSet},
+    fs,
+    io::Write,
+    process::Command,
+    rc::Rc,
+    sync::Arc,
+};
 
-use inkwell::{
+use itertools::Itertools;
+use parking_lot::{Mutex, RwLock};
+use pyo3::{
+    create_exception, exceptions,
+    prelude::*,
+    types::{PyBytes, PyDict, PyNone, PySet},
+};
+use tempfile::{self, TempDir};
+
+use nac3core::{
+    codegen::{
+        concrete_type::ConcreteTypeStore, gen_func_impl, irrt::load_irrt, CodeGenLLVMOptions,
+        CodeGenTargetMachineOptions, CodeGenTask, CodeGenerator, WithCall, WorkerRegistry,
+    },
+    inkwell::{
+        context::Context,
         memory_buffer::MemoryBuffer,
-    module::{Linkage, Module},
+        module::{FlagBehavior, Linkage, Module},
         passes::PassBuilderOptions,
         support::is_multithreaded,
         targets::*,
         OptimizationLevel,
-};
-use itertools::Itertools;
-use nac3core::codegen::{gen_func_impl, CodeGenLLVMOptions, CodeGenTargetMachineOptions};
-use nac3core::toplevel::builtins::get_exn_constructor;
-use nac3core::typecheck::typedef::{TypeEnum, Unifier, VarMap};
-use nac3parser::{
-    ast::{ExprKind, Stmt, StmtKind, StrRef},
+    },
+    nac3parser::{
+        ast::{Constant, ExprKind, Located, Stmt, StmtKind, StrRef},
         parser::parse_program,
-};
-use pyo3::create_exception;
-use pyo3::prelude::*;
-use pyo3::{exceptions, types::PyBytes, types::PyDict, types::PySet};
-
-use parking_lot::{Mutex, RwLock};
-
-use nac3core::{
-    codegen::irrt::load_irrt,
-    codegen::{concrete_type::ConcreteTypeStore, CodeGenTask, WithCall, WorkerRegistry},
+    },
     symbol_resolver::SymbolResolver,
     toplevel::{
-        composer::{ComposerConfig, TopLevelComposer},
+        builtins::get_exn_constructor,
+        composer::{BuiltinFuncCreator, BuiltinFuncSpec, ComposerConfig, TopLevelComposer},
         DefinitionId, GenCall, TopLevelDef,
     },
-    typecheck::typedef::{FunSignature, FuncArg},
-    typecheck::{type_inferencer::PrimitiveStore, typedef::Type},
+    typecheck::{
+        type_inferencer::PrimitiveStore,
+        typedef::{into_var_map, FunSignature, FuncArg, Type, TypeEnum, Unifier, VarMap},
+    },
 };
-
 use nac3ld::Linker;
 
-use tempfile::{self, TempDir};
-
-use crate::codegen::attributes_writeback;
-use crate::{
-    codegen::{rpc_codegen_callback, ArtiqCodeGenerator},
-    symbol_resolver::{DeferredEvaluationStore, InnerResolver, PythonHelper, Resolver},
+use codegen::{
+    attributes_writeback, gen_core_log, gen_rtio_log, rpc_codegen_callback, ArtiqCodeGenerator,
 };
+use symbol_resolver::{DeferredEvaluationStore, InnerResolver, PythonHelper, Resolver};
+use timeline::TimeFns;
 
 mod codegen;
 mod symbol_resolver;
 mod timeline;
 
-use timeline::TimeFns;
-
 #[derive(PartialEq, Clone, Copy)]
 enum Isa {
     Host,
@@ -126,7 +122,7 @@ struct Nac3 {
     isa: Isa,
     time_fns: &'static (dyn TimeFns + Sync),
     primitive: PrimitiveStore,
-    builtins: Vec<(StrRef, FunSignature, Arc<GenCall>)>,
+    builtins: Vec<BuiltinFuncSpec>,
     pyid_to_def: Arc<RwLock<HashMap<u64, DefinitionId>>>,
     primitive_ids: PrimitivePythonId,
     working_directory: TempDir,
@@ -146,14 +142,32 @@ impl Nac3 {
         module: &PyObject,
         registered_class_ids: &HashSet<u64>,
     ) -> PyResult<()> {
-        let (module_name, source_file) = Python::with_gil(|py| -> PyResult<(String, String)> {
+        let (module_name, source_file, source) =
+            Python::with_gil(|py| -> PyResult<(String, String, String)> {
                 let module: &PyAny = module.extract(py)?;
-            Ok((module.getattr("__name__")?.extract()?, module.getattr("__file__")?.extract()?))
+                let source_file = module.getattr("__file__");
+                let (source_file, source) = if let Ok(source_file) = source_file {
+                    let source_file = source_file.extract()?;
+                    (
+                        source_file,
+                        fs::read_to_string(&source_file).map_err(|e| {
+                            exceptions::PyIOError::new_err(format!(
+                                "failed to read input file: {e}"
+                            ))
+                        })?,
+                    )
+                } else {
+                    // kernels submitted by content have no file
+                    // but still can provide source by StringLoader
+                    let get_src_fn = module
+                        .getattr("__loader__")?
+                        .extract::<PyObject>()?
+                        .getattr(py, "get_source")?;
+                    ("<expcontent>", get_src_fn.call1(py, (PyNone::get(py),))?.extract(py)?)
+                };
+                Ok((module.getattr("__name__")?.extract()?, source_file.to_string(), source))
             })?;
 
-        let source = fs::read_to_string(&source_file).map_err(|e| {
-            exceptions::PyIOError::new_err(format!("failed to read input file: {e}"))
-        })?;
         let parser_result = parse_program(&source, source_file.into())
             .map_err(|e| exceptions::PySyntaxError::new_err(format!("parse error: {e}")))?;
 
@@ -193,10 +207,8 @@ impl Nac3 {
                     body.retain(|stmt| {
                         if let StmtKind::FunctionDef { ref decorator_list, .. } = stmt.node {
                             decorator_list.iter().any(|decorator| {
-                                if let ExprKind::Name { id, .. } = decorator.node {
-                                    id.to_string() == "kernel"
-                                        || id.to_string() == "portable"
-                                        || id.to_string() == "rpc"
+                                if let Some(id) = decorator_id_string(decorator) {
+                                    id == "kernel" || id == "portable" || id == "rpc"
                                 } else {
                                     false
                                 }
@@ -209,9 +221,8 @@ impl Nac3 {
                 }
                 StmtKind::FunctionDef { ref decorator_list, .. } => {
                     decorator_list.iter().any(|decorator| {
-                        if let ExprKind::Name { id, .. } = decorator.node {
-                            let id = id.to_string();
-                            id == "extern" || id == "portable" || id == "kernel" || id == "rpc"
+                        if let Some(id) = decorator_id_string(decorator) {
+                            id == "extern" || id == "kernel" || id == "portable" || id == "rpc"
                         } else {
                             false
                         }
@@ -264,7 +275,7 @@ impl Nac3 {
                     arg_names.len(),
                 ));
             }
-            for (i, FuncArg { ty, default_value, name }) in args.iter().enumerate() {
+            for (i, FuncArg { ty, default_value, name, .. }) in args.iter().enumerate() {
                 let in_name = match arg_names.get(i) {
                     Some(n) => n,
                     None if default_value.is_none() => {
@@ -300,6 +311,64 @@ impl Nac3 {
         None
     }
 
+    /// Returns a [`Vec`] of builtins that needs to be initialized during method compilation time.
+    fn get_lateinit_builtins() -> Vec<Box<BuiltinFuncCreator>> {
+        vec![
+            Box::new(|primitives, unifier| {
+                let arg_ty = unifier.get_fresh_var(Some("T".into()), None);
+
+                (
+                    "core_log".into(),
+                    FunSignature {
+                        args: vec![FuncArg {
+                            name: "arg".into(),
+                            ty: arg_ty.ty,
+                            default_value: None,
+                            is_vararg: false,
+                        }],
+                        ret: primitives.none,
+                        vars: into_var_map([arg_ty]),
+                    },
+                    Arc::new(GenCall::new(Box::new(move |ctx, obj, fun, args, generator| {
+                        gen_core_log(ctx, &obj, fun, &args, generator)?;
+
+                        Ok(None)
+                    }))),
+                )
+            }),
+            Box::new(|primitives, unifier| {
+                let arg_ty = unifier.get_fresh_var(Some("T".into()), None);
+
+                (
+                    "rtio_log".into(),
+                    FunSignature {
+                        args: vec![
+                            FuncArg {
+                                name: "channel".into(),
+                                ty: primitives.str,
+                                default_value: None,
+                                is_vararg: false,
+                            },
+                            FuncArg {
+                                name: "arg".into(),
+                                ty: arg_ty.ty,
+                                default_value: None,
+                                is_vararg: false,
+                            },
+                        ],
+                        ret: primitives.none,
+                        vars: into_var_map([arg_ty]),
+                    },
+                    Arc::new(GenCall::new(Box::new(move |ctx, obj, fun, args, generator| {
+                        gen_rtio_log(ctx, &obj, fun, &args, generator)?;
+
+                        Ok(None)
+                    }))),
+                )
+            }),
+        ]
+    }
+
     fn compile_method<T>(
         &self,
         obj: &PyAny,
@@ -312,6 +381,7 @@ impl Nac3 {
         let size_t = self.isa.get_size_type();
         let (mut composer, mut builtins_def, mut builtins_ty) = TopLevelComposer::new(
             self.builtins.clone(),
+            Self::get_lateinit_builtins(),
             ComposerConfig { kernel_ann: Some("Kernel"), kernel_invariant_ann: "KernelInvariant" },
             size_t,
         );
@@ -388,7 +458,6 @@ impl Nac3 {
                         pyid_to_type: pyid_to_type.clone(),
                         primitive_ids: self.primitive_ids.clone(),
                         global_value_ids: global_value_ids.clone(),
-                        class_names: Mutex::default(),
                         name_to_pyid: name_to_pyid.clone(),
                         module: module.clone(),
                         id_to_pyval: RwLock::default(),
@@ -419,9 +488,25 @@ impl Nac3 {
 
             match &stmt.node {
                 StmtKind::FunctionDef { decorator_list, .. } => {
-                    if decorator_list.iter().any(|decorator| matches!(decorator.node, ExprKind::Name { id, .. } if id == "rpc".into())) {
-                        store_fun.call1(py, (def_id.0.into_py(py), module.getattr(py, name.to_string().as_str()).unwrap())).unwrap();
-                        rpc_ids.push((None, def_id));
+                    if decorator_list
+                        .iter()
+                        .any(|decorator| decorator_id_string(decorator) == Some("rpc".to_string()))
+                    {
+                        store_fun
+                            .call1(
+                                py,
+                                (
+                                    def_id.0.into_py(py),
+                                    module.getattr(py, name.to_string().as_str()).unwrap(),
+                                ),
+                            )
+                            .unwrap();
+                        let is_async = decorator_list.iter().any(|decorator| {
+                            decorator_get_flags(decorator)
+                                .iter()
+                                .any(|constant| *constant == Constant::Str("async".into()))
+                        });
+                        rpc_ids.push((None, def_id, is_async));
                     }
                 }
                 StmtKind::ClassDef { name, body, .. } => {
@@ -429,19 +514,26 @@ impl Nac3 {
                     let class_obj = module.getattr(py, class_name.as_str()).unwrap();
                     for stmt in body {
                         if let StmtKind::FunctionDef { name, decorator_list, .. } = &stmt.node {
-                            if decorator_list.iter().any(|decorator| matches!(decorator.node, ExprKind::Name { id, .. } if id == "rpc".into())) {
+                            if decorator_list.iter().any(|decorator| {
+                                decorator_id_string(decorator) == Some("rpc".to_string())
+                            }) {
+                                let is_async = decorator_list.iter().any(|decorator| {
+                                    decorator_get_flags(decorator)
+                                        .iter()
+                                        .any(|constant| *constant == Constant::Str("async".into()))
+                                });
                                 if name == &"__init__".into() {
                                     return Err(CompileError::new_err(format!(
                                         "compilation failed\n----------\nThe constructor of class {} should not be decorated with rpc decorator (at {})",
                                         class_name, stmt.location
                                     )));
                                 }
-                                rpc_ids.push((Some((class_obj.clone(), *name)), def_id));
+                                rpc_ids.push((Some((class_obj.clone(), *name)), def_id, is_async));
                             }
                         }
                     }
                 }
-                _ => ()
+                _ => (),
             }
 
             let id = *name_to_pyid.get(&name).unwrap();
@@ -480,7 +572,6 @@ impl Nac3 {
             pyid_to_type: pyid_to_type.clone(),
             primitive_ids: self.primitive_ids.clone(),
             global_value_ids: global_value_ids.clone(),
-            class_names: Mutex::default(),
             id_to_pyval: RwLock::default(),
             id_to_primitive: RwLock::default(),
             field_to_val: RwLock::default(),
@@ -497,6 +588,10 @@ impl Nac3 {
             .register_top_level(synthesized.pop().unwrap(), Some(resolver.clone()), "", false)
             .unwrap();
 
+        // Process IRRT
+        let context = Context::create();
+        let irrt = load_irrt(&context, resolver.as_ref());
+
         let fun_signature =
             FunSignature { args: vec![], ret: self.primitive.none, vars: VarMap::new() };
         let mut store = ConcreteTypeStore::new();
@@ -534,13 +629,12 @@ impl Nac3 {
         let top_level = Arc::new(composer.make_top_level_context());
 
         {
-            let rpc_codegen = rpc_codegen_callback();
             let defs = top_level.definitions.read();
-            for (class_data, id) in &rpc_ids {
+            for (class_data, id, is_async) in &rpc_ids {
                 let mut def = defs[id.0].write();
                 match &mut *def {
                     TopLevelDef::Function { codegen_callback, .. } => {
-                        *codegen_callback = Some(rpc_codegen.clone());
+                        *codegen_callback = Some(rpc_codegen_callback(*is_async));
                     }
                     TopLevelDef::Class { methods, .. } => {
                         let (class_def, method_name) = class_data.as_ref().unwrap();
@@ -551,7 +645,7 @@ impl Nac3 {
                             if let TopLevelDef::Function { codegen_callback, .. } =
                                 &mut *defs[id.0].write()
                             {
-                                *codegen_callback = Some(rpc_codegen.clone());
+                                *codegen_callback = Some(rpc_codegen_callback(*is_async));
                                 store_fun
                                     .call1(
                                         py,
@@ -566,6 +660,11 @@ impl Nac3 {
                             }
                         }
                     }
+                    TopLevelDef::Variable { .. } => {
+                        return Err(CompileError::new_err(String::from(
+                            "Unsupported @rpc annotation on global variable",
+                        )))
+                    }
                 }
             }
         }
@@ -586,33 +685,12 @@ impl Nac3 {
         let task = CodeGenTask {
             subst: Vec::default(),
             symbol_name: "__modinit__".to_string(),
-            body: instance.body,
-            signature,
-            resolver: resolver.clone(),
-            store,
-            unifier_index: instance.unifier_id,
-            calls: instance.calls,
-            id: 0,
-        };
-
-        let mut store = ConcreteTypeStore::new();
-        let mut cache = HashMap::new();
-        let signature = store.from_signature(
-            &mut composer.unifier,
-            &self.primitive,
-            &fun_signature,
-            &mut cache,
-        );
-        let signature = store.add_cty(signature);
-        let attributes_writeback_task = CodeGenTask {
-            subst: Vec::default(),
-            symbol_name: "attributes_writeback".to_string(),
             body: Arc::new(Vec::default()),
             signature,
             resolver,
             store,
             unifier_index: instance.unifier_id,
-            calls: Arc::new(HashMap::default()),
+            calls: instance.calls,
             id: 0,
         };
 
@@ -625,7 +703,9 @@ impl Nac3 {
             let buffer = buffer.as_slice().into();
             membuffer.lock().push(buffer);
         })));
-        let size_t = if self.isa == Isa::Host { 64 } else { 32 };
+        let size_t = context
+            .ptr_sized_int_type(&self.get_llvm_target_machine().get_target_data(), None)
+            .get_bit_width();
         let num_threads = if is_multithreaded() { 4 } else { 1 };
         let thread_names: Vec<String> = (0..num_threads).map(|_| "main".to_string()).collect();
         let threads: Vec<_> = thread_names
@@ -634,16 +714,27 @@ impl Nac3 {
             .collect();
 
         let membuffer = membuffers.clone();
+        let mut has_return = false;
         py.allow_threads(|| {
             let (registry, handles) =
                 WorkerRegistry::create_workers(threads, top_level.clone(), &self.llvm_options, &f);
-            registry.add_task(task);
-            registry.wait_tasks_complete(handles);
 
-            let mut generator =
-                ArtiqCodeGenerator::new("attributes_writeback".to_string(), size_t, self.time_fns);
-            let context = inkwell::context::Context::create();
-            let module = context.create_module("attributes_writeback");
+            let mut generator = ArtiqCodeGenerator::new("main".to_string(), size_t, self.time_fns);
+            let context = Context::create();
+            let module = context.create_module("main");
+            let target_machine = self.llvm_options.create_target_machine().unwrap();
+            module.set_data_layout(&target_machine.get_target_data().get_data_layout());
+            module.set_triple(&target_machine.get_triple());
+            module.add_basic_value_flag(
+                "Debug Info Version",
+                FlagBehavior::Warning,
+                context.i32_type().const_int(3, false),
+            );
+            module.add_basic_value_flag(
+                "Dwarf Version",
+                FlagBehavior::Warning,
+                context.i32_type().const_int(4, false),
+            );
             let builder = context.create_builder();
             let (_, module, _) = gen_func_impl(
                 &context,
@@ -651,9 +742,27 @@ impl Nac3 {
                 &registry,
                 builder,
                 module,
-                attributes_writeback_task,
+                task,
                 |generator, ctx| {
-                    attributes_writeback(ctx, generator, inner_resolver.as_ref(), &host_attributes)
+                    assert_eq!(instance.body.len(), 1, "toplevel module should have 1 statement");
+                    let StmtKind::Expr { value: ref expr, .. } = instance.body[0].node else {
+                        unreachable!("toplevel statement must be an expression")
+                    };
+                    let ExprKind::Call { .. } = expr.node else {
+                        unreachable!("toplevel expression must be a function call")
+                    };
+
+                    let return_obj =
+                        generator.gen_expr(ctx, expr)?.map(|value| (expr.custom.unwrap(), value));
+                    has_return = return_obj.is_some();
+                    registry.wait_tasks_complete(handles);
+                    attributes_writeback(
+                        ctx,
+                        generator,
+                        inner_resolver.as_ref(),
+                        &host_attributes,
+                        return_obj,
+                    )
                 },
             )
             .unwrap();
@@ -662,37 +771,24 @@ impl Nac3 {
             membuffer.lock().push(buffer);
         });
 
-        let context = inkwell::context::Context::create();
+        embedding_map.setattr("expects_return", has_return).unwrap();
+
+        // Link all modules into `main`.
         let buffers = membuffers.lock();
         let main = context
-            .create_module_from_ir(MemoryBuffer::create_from_memory_range(&buffers[0], "main"))
+            .create_module_from_ir(MemoryBuffer::create_from_memory_range(
+                buffers.last().unwrap(),
+                "main",
+            ))
             .unwrap();
-        for buffer in buffers.iter().skip(1) {
+        for buffer in buffers.iter().rev().skip(1) {
             let other = context
                 .create_module_from_ir(MemoryBuffer::create_from_memory_range(buffer, "main"))
                 .unwrap();
 
             main.link_in_module(other).map_err(|err| CompileError::new_err(err.to_string()))?;
         }
-        let builder = context.create_builder();
-        let modinit_return = main
-            .get_function("__modinit__")
-            .unwrap()
-            .get_last_basic_block()
-            .unwrap()
-            .get_terminator()
-            .unwrap();
-        builder.position_before(&modinit_return);
-        builder
-            .build_call(
-                main.get_function("attributes_writeback").unwrap(),
-                &[],
-                "attributes_writeback",
-            )
-            .unwrap();
-
-        main.link_in_module(load_irrt(&context))
-            .map_err(|err| CompileError::new_err(err.to_string()))?;
+        main.link_in_module(irrt).map_err(|err| CompileError::new_err(err.to_string()))?;
 
         let mut function_iter = main.get_first_function();
         while let Some(func) = function_iter {
@@ -778,6 +874,41 @@ impl Nac3 {
     }
 }
 
+/// Retrieves the Name.id from a decorator, supports decorators with arguments.
+fn decorator_id_string(decorator: &Located<ExprKind>) -> Option<String> {
+    if let ExprKind::Name { id, .. } = decorator.node {
+        // Bare decorator
+        return Some(id.to_string());
+    } else if let ExprKind::Call { func, .. } = &decorator.node {
+        // Decorators that are calls (e.g. "@rpc()") have Call for the node,
+        // need to extract the id from within.
+        if let ExprKind::Name { id, .. } = func.node {
+            return Some(id.to_string());
+        }
+    }
+    None
+}
+
+/// Retrieves flags from a decorator, if any.
+fn decorator_get_flags(decorator: &Located<ExprKind>) -> Vec<Constant> {
+    let mut flags = vec![];
+    if let ExprKind::Call { keywords, .. } = &decorator.node {
+        for keyword in keywords {
+            if keyword.node.arg != Some("flags".into()) {
+                continue;
+            }
+            if let ExprKind::Set { elts } = &keyword.node.value.node {
+                for elt in elts {
+                    if let ExprKind::Constant { value, .. } = &elt.node {
+                        flags.push(value.clone());
+                    }
+                }
+            }
+        }
+    }
+    flags
+}
+
 fn link_with_lld(elf_filename: String, obj_filename: String) -> PyResult<()> {
     let linker_args = vec![
         "-shared".to_string(),
@@ -847,7 +978,7 @@ impl Nac3 {
             Isa::RiscV32IMA => &timeline::NOW_PINNING_TIME_FNS,
             Isa::CortexA9 | Isa::Host => &timeline::EXTERN_TIME_FNS,
         };
-        let primitive: PrimitiveStore = TopLevelComposer::make_primitives(isa.get_size_type()).0;
+        let (primitive, _) = TopLevelComposer::make_primitives(isa.get_size_type());
         let builtins = vec![
             (
                 "now_mu".into(),
@@ -863,6 +994,7 @@ impl Nac3 {
                         name: "t".into(),
                         ty: primitive.int64,
                         default_value: None,
+                        is_vararg: false,
                     }],
                     ret: primitive.none,
                     vars: VarMap::new(),
@@ -882,6 +1014,7 @@ impl Nac3 {
                         name: "dt".into(),
                         ty: primitive.int64,
                         default_value: None,
+                        is_vararg: false,
                     }],
                     ret: primitive.none,
                     vars: VarMap::new(),
@@ -957,7 +1090,12 @@ impl Nac3 {
         })
     }
 
-    fn analyze(&mut self, functions: &PySet, classes: &PySet) -> PyResult<()> {
+    fn analyze(
+        &mut self,
+        functions: &PySet,
+        classes: &PySet,
+        content_modules: &PySet,
+    ) -> PyResult<()> {
         let (modules, class_ids) =
             Python::with_gil(|py| -> PyResult<(HashMap<u64, PyObject>, HashSet<u64>)> {
                 let mut modules: HashMap<u64, PyObject> = HashMap::new();
@@ -967,14 +1105,22 @@ impl Nac3 {
                 let getmodule_fn = PyModule::import(py, "inspect")?.getattr("getmodule")?;
 
                 for function in functions {
-                    let module = getmodule_fn.call1((function,))?.extract()?;
+                    let module: PyObject = getmodule_fn.call1((function,))?.extract()?;
+                    if !module.is_none(py) {
                         modules.insert(id_fn.call1((&module,))?.extract()?, module);
                     }
+                }
                 for class in classes {
-                    let module = getmodule_fn.call1((class,))?.extract()?;
+                    let module: PyObject = getmodule_fn.call1((class,))?.extract()?;
+                    if !module.is_none(py) {
                         modules.insert(id_fn.call1((&module,))?.extract()?, module);
+                    }
                     class_ids.insert(id_fn.call1((class,))?.extract()?);
                 }
+                for module in content_modules {
+                    let module: PyObject = module.extract()?;
+                    modules.insert(id_fn.call1((&module,))?.extract()?, module.into());
+                }
                 Ok((modules, class_ids))
             })?;
 

nac3artiq/src/symbol_resolver.rs

@@ -1,14 +1,27 @@
-use inkwell::{
+use std::{
+    collections::{HashMap, HashSet},
+    sync::{
+        atomic::{AtomicBool, Ordering::Relaxed},
+        Arc,
+    },
+};
+
+use itertools::Itertools;
+use parking_lot::RwLock;
+use pyo3::{
+    types::{PyDict, PyTuple},
+    PyAny, PyObject, PyResult, Python,
+};
+
+use nac3core::{
+    codegen::{types::NDArrayType, CodeGenContext, CodeGenerator},
+    inkwell::{
+        module::Linkage,
         types::{BasicType, BasicTypeEnum},
         values::BasicValueEnum,
         AddressSpace,
-};
-use itertools::Itertools;
-use nac3core::{
-    codegen::{
-        classes::{NDArrayType, ProxyType},
-        CodeGenContext, CodeGenerator,
     },
+    nac3parser::ast::{self, StrRef},
     symbol_resolver::{StaticValue, SymbolResolver, SymbolValue, ValueEnum},
     toplevel::{
         helper::PrimDef,
@@ -20,21 +33,8 @@ use nac3core::{
         typedef::{into_var_map, iter_type_vars, Type, TypeEnum, TypeVar, Unifier, VarMap},
     },
 };
-use nac3parser::ast::{self, StrRef};
-use parking_lot::{Mutex, RwLock};
-use pyo3::{
-    types::{PyDict, PyTuple},
-    PyAny, PyObject, PyResult, Python,
-};
-use std::{
-    collections::{HashMap, HashSet},
-    sync::{
-        atomic::{AtomicBool, Ordering::Relaxed},
-        Arc,
-    },
-};
 
-use crate::PrimitivePythonId;
+use super::PrimitivePythonId;
 
 pub enum PrimitiveValue {
     I32(i32),
@@ -79,7 +79,6 @@ pub struct InnerResolver {
     pub id_to_primitive: RwLock<HashMap<u64, PrimitiveValue>>,
     pub field_to_val: RwLock<HashMap<ResolverField, Option<PyFieldHandle>>>,
     pub global_value_ids: Arc<RwLock<HashMap<u64, PyObject>>>,
-    pub class_names: Mutex<HashMap<StrRef, Type>>,
     pub pyid_to_def: Arc<RwLock<HashMap<u64, DefinitionId>>>,
     pub pyid_to_type: Arc<RwLock<HashMap<u64, Type>>>,
     pub primitive_ids: PrimitivePythonId,
@@ -133,6 +132,8 @@ impl StaticValue for PythonValue {
                         format!("{}_const", self.id).as_str(),
                     );
                     global.set_constant(true);
+                    // Set linkage of global to private to avoid name collisions
+                    global.set_linkage(Linkage::Private);
                     global.set_initializer(&ctx.ctx.const_struct(
                         &[ctx.ctx.i32_type().const_int(u64::from(id), false).into()],
                         false,
@@ -163,7 +164,7 @@ impl StaticValue for PythonValue {
                 PrimitiveValue::Bool(val) => {
                     ctx.ctx.i8_type().const_int(u64::from(*val), false).into()
                 }
-                PrimitiveValue::Str(val) => ctx.ctx.const_string(val.as_bytes(), true).into(),
+                PrimitiveValue::Str(val) => ctx.gen_string(generator, val).into(),
             });
         }
         if let Some(global) = ctx.module.get_global(&self.id.to_string()) {
@@ -351,7 +352,7 @@ impl InnerResolver {
             Ok(Ok((ndarray, false)))
         } else if ty_id == self.primitive_ids.tuple {
             // do not handle type var param and concrete check here
-            Ok(Ok((unifier.add_ty(TypeEnum::TTuple { ty: vec![] }), false)))
+            Ok(Ok((unifier.add_ty(TypeEnum::TTuple { ty: vec![], is_vararg_ctx: false }), false)))
         } else if ty_id == self.primitive_ids.option {
             Ok(Ok((primitives.option, false)))
         } else if ty_id == self.primitive_ids.none {
@@ -555,7 +556,10 @@ impl InnerResolver {
                             Err(err) => return Ok(Err(err)),
                             _ => return Ok(Err("tuple type needs at least 1 type parameters".to_string()))
                         };
-                    Ok(Ok((unifier.add_ty(TypeEnum::TTuple { ty: args }), true)))
+                    Ok(Ok((
+                        unifier.add_ty(TypeEnum::TTuple { ty: args, is_vararg_ctx: false }),
+                        true,
+                    )))
                 }
                 TypeEnum::TObj { params, obj_id, .. } => {
                     let subst = {
@@ -797,7 +801,9 @@ impl InnerResolver {
                     .map(|elem| self.get_obj_type(py, elem, unifier, defs, primitives))
                     .collect();
                 let types = types?;
-                Ok(types.map(|types| unifier.add_ty(TypeEnum::TTuple { ty: types })))
+                Ok(types.map(|types| {
+                    unifier.add_ty(TypeEnum::TTuple { ty: types, is_vararg_ctx: false })
+                }))
             }
             // special handling for option type since its class member layout in python side
             // is special and cannot be mapped directly to a nac3 type as below
@@ -972,7 +978,7 @@ impl InnerResolver {
         } else if ty_id == self.primitive_ids.string || ty_id == self.primitive_ids.np_str_ {
             let val: String = obj.extract().unwrap();
             self.id_to_primitive.write().insert(id, PrimitiveValue::Str(val.clone()));
-            Ok(Some(ctx.ctx.const_string(val.as_bytes(), true).into()))
+            Ok(Some(ctx.gen_string(generator, val).into()))
         } else if ty_id == self.primitive_ids.float || ty_id == self.primitive_ids.float64 {
             let val: f64 = obj.extract().unwrap();
             self.id_to_primitive.write().insert(id, PrimitiveValue::F64(val));
@@ -991,8 +997,15 @@ impl InnerResolver {
                 }
                 _ => unreachable!("must be list"),
             };
-            let ty = ctx.get_llvm_type(generator, elem_ty);
             let size_t = generator.get_size_type(ctx.ctx);
+            let ty = if len == 0
+                && matches!(&*ctx.unifier.get_ty_immutable(elem_ty), TypeEnum::TVar { .. })
+            {
+                // The default type for zero-length lists of unknown element type is size_t
+                size_t.into()
+            } else {
+                ctx.get_llvm_type(generator, elem_ty)
+            };
             let arr_ty = ctx
                 .ctx
                 .struct_type(&[ty.ptr_type(AddressSpace::default()).into(), size_t.into()], false);
@@ -1080,7 +1093,7 @@ impl InnerResolver {
                 if self.global_value_ids.read().contains_key(&id) {
                     let global = ctx.module.get_global(&id_str).unwrap_or_else(|| {
                         ctx.module.add_global(
-                            ndarray_llvm_ty.as_underlying_type(),
+                            ndarray_llvm_ty.element_type().into_struct_type(),
                             Some(AddressSpace::default()),
                             &id_str,
                         )
@@ -1174,7 +1187,7 @@ impl InnerResolver {
             data_global.set_initializer(&data);
 
             // create a global for the ndarray object and initialize it
-            let value = ndarray_llvm_ty.as_underlying_type().const_named_struct(&[
+            let value = ndarray_llvm_ty.element_type().into_struct_type().const_named_struct(&[
                 llvm_usize.const_int(ndarray_ndims, false).into(),
                 shape_global
                     .as_pointer_value()
@@ -1187,7 +1200,7 @@ impl InnerResolver {
             ]);
 
             let ndarray = ctx.module.add_global(
-                ndarray_llvm_ty.as_underlying_type(),
+                ndarray_llvm_ty.element_type().into_struct_type(),
                 Some(AddressSpace::default()),
                 &id_str,
             );
@@ -1196,7 +1209,9 @@ impl InnerResolver {
             Ok(Some(ndarray.as_pointer_value().into()))
         } else if ty_id == self.primitive_ids.tuple {
             let expected_ty_enum = ctx.unifier.get_ty_immutable(expected_ty);
-            let TypeEnum::TTuple { ty } = expected_ty_enum.as_ref() else { unreachable!() };
+            let TypeEnum::TTuple { ty, is_vararg_ctx: false } = expected_ty_enum.as_ref() else {
+                unreachable!()
+            };
 
             let tup_tys = ty.iter();
             let elements: &PyTuple = obj.downcast()?;
@@ -1452,6 +1467,7 @@ impl SymbolResolver for Resolver {
         &self,
         id: StrRef,
         _: &mut CodeGenContext<'ctx, '_>,
+        _: &mut dyn CodeGenerator,
     ) -> Option<ValueEnum<'ctx>> {
         let sym_value = {
             let id_to_val = self.0.id_to_pyval.read();

nac3artiq/src/timeline.rs

@@ -1,9 +1,12 @@
-use inkwell::{
+use itertools::Either;
+
+use nac3core::{
+    codegen::CodeGenContext,
+    inkwell::{
         values::{BasicValueEnum, CallSiteValue},
         AddressSpace, AtomicOrdering,
+    },
 };
-use itertools::Either;
-use nac3core::codegen::CodeGenContext;
 
 /// Functions for manipulating the timeline.
 pub trait TimeFns {
@@ -31,7 +34,7 @@ impl TimeFns for NowPinningTimeFns64 {
             .unwrap_or_else(|| ctx.module.add_global(i64_type, None, "now"));
         let now_hiptr = ctx
             .builder
-            .build_bitcast(now, i32_type.ptr_type(AddressSpace::default()), "now.hi.addr")
+            .build_bit_cast(now, i32_type.ptr_type(AddressSpace::default()), "now.hi.addr")
             .map(BasicValueEnum::into_pointer_value)
             .unwrap();
 
@@ -80,7 +83,7 @@ impl TimeFns for NowPinningTimeFns64 {
             .unwrap_or_else(|| ctx.module.add_global(i64_type, None, "now"));
         let now_hiptr = ctx
             .builder
-            .build_bitcast(now, i32_type.ptr_type(AddressSpace::default()), "now.hi.addr")
+            .build_bit_cast(now, i32_type.ptr_type(AddressSpace::default()), "now.hi.addr")
             .map(BasicValueEnum::into_pointer_value)
             .unwrap();
 
@@ -109,7 +112,7 @@ impl TimeFns for NowPinningTimeFns64 {
             .unwrap_or_else(|| ctx.module.add_global(i64_type, None, "now"));
         let now_hiptr = ctx
             .builder
-            .build_bitcast(now, i32_type.ptr_type(AddressSpace::default()), "now.hi.addr")
+            .build_bit_cast(now, i32_type.ptr_type(AddressSpace::default()), "now.hi.addr")
             .map(BasicValueEnum::into_pointer_value)
             .unwrap();
 
@@ -207,7 +210,7 @@ impl TimeFns for NowPinningTimeFns {
             .unwrap_or_else(|| ctx.module.add_global(i64_type, None, "now"));
         let now_hiptr = ctx
             .builder
-            .build_bitcast(now, i32_type.ptr_type(AddressSpace::default()), "now.hi.addr")
+            .build_bit_cast(now, i32_type.ptr_type(AddressSpace::default()), "now.hi.addr")
             .map(BasicValueEnum::into_pointer_value)
             .unwrap();
 
@@ -258,7 +261,7 @@ impl TimeFns for NowPinningTimeFns {
         let time_lo = ctx.builder.build_int_truncate(time, i32_type, "time.lo").unwrap();
         let now_hiptr = ctx
             .builder
-            .build_bitcast(now, i32_type.ptr_type(AddressSpace::default()), "now.hi.addr")
+            .build_bit_cast(now, i32_type.ptr_type(AddressSpace::default()), "now.hi.addr")
             .map(BasicValueEnum::into_pointer_value)
             .unwrap();
 

nac3ast/Cargo.toml

@@ -10,7 +10,6 @@ constant-optimization = ["fold"]
 fold = []
 
 [dependencies]
-lazy_static = "1.5"
 parking_lot = "0.12"
 string-interner = "0.17"
 fxhash = "0.2"

nac3ast/src/ast_gen.rs

@@ -5,14 +5,12 @@ pub use crate::location::Location;
 
 use fxhash::FxBuildHasher;
 use parking_lot::{Mutex, MutexGuard};
-use std::{cell::RefCell, collections::HashMap, fmt};
+use std::{cell::RefCell, collections::HashMap, fmt, sync::LazyLock};
 use string_interner::{symbol::SymbolU32, DefaultBackend, StringInterner};
 
 pub type Interner = StringInterner<DefaultBackend, FxBuildHasher>;
-lazy_static! {
-    static ref INTERNER: Mutex<Interner> =
-        Mutex::new(StringInterner::with_hasher(FxBuildHasher::default()));
-}
+static INTERNER: LazyLock<Mutex<Interner>> =
+    LazyLock::new(|| Mutex::new(StringInterner::with_hasher(FxBuildHasher::default())));
 
 thread_local! {
     static LOCAL_INTERNER: RefCell<HashMap<String, StrRef>> = RefCell::default();

nac3ast/src/lib.rs

@@ -1,10 +1,4 @@
-#![deny(
-    future_incompatible,
-    let_underscore,
-    nonstandard_style,
-    rust_2024_compatibility,
-    clippy::all
-)]
+#![deny(future_incompatible, let_underscore, nonstandard_style, clippy::all)]
 #![warn(clippy::pedantic)]
 #![allow(
     clippy::missing_errors_doc,
@@ -14,9 +8,6 @@
     clippy::wildcard_imports
 )]
 
-#[macro_use]
-extern crate lazy_static;
-
 mod ast_gen;
 mod constant;
 #[cfg(feature = "fold")]

nac3core/Cargo.toml

@@ -1,26 +1,29 @@
-[features]
-test = []
-
 [package]
 name = "nac3core"
 version = "0.1.0"
 authors = ["M-Labs"]
 edition = "2021"
 
+[features]
+default = ["derive"]
+derive = ["dep:nac3core_derive"]
+no-escape-analysis = []
+
 [dependencies]
 itertools = "0.13"
 crossbeam = "0.8"
-indexmap = "2.2"
+indexmap = "2.6"
 parking_lot = "0.12"
-rayon = "1.8"
+rayon = "1.10"
+nac3core_derive = { path = "nac3core_derive", optional = true }
 nac3parser = { path = "../nac3parser" }
-strum = "0.26.2"
-strum_macros = "0.26.4"
+strum = "0.26"
+strum_macros = "0.26"
 
 [dependencies.inkwell]
-version = "0.4"
+version = "0.5"
 default-features = false
-features = ["llvm14-0", "target-x86", "target-arm", "target-riscv", "no-libffi-linking"]
+features = ["llvm14-0-prefer-dynamic", "target-x86", "target-arm", "target-riscv", "no-libffi-linking"]
 
 [dev-dependencies]
 test-case = "1.2.0"

nac3core/build.rs

@@ -1,4 +1,3 @@
-use regex::Regex;
 use std::{
     env,
     fs::File,
@@ -7,45 +6,58 @@ use std::{
     process::{Command, Stdio},
 };
 
-fn compile_irrt(irrt_dir: &Path, out_dir: &Path) {
+use regex::Regex;
+
+fn main() {
+    let out_dir = env::var("OUT_DIR").unwrap();
+    let out_dir = Path::new(&out_dir);
+    let irrt_dir = Path::new("irrt");
+
     let irrt_cpp_path = irrt_dir.join("irrt.cpp");
 
     /*
      * HACK: Sadly, clang doesn't let us emit generic LLVM bitcode.
      * Compiling for WASM32 and filtering the output with regex is the closest we can get.
      */
-    let flags: &[&str] = &[
+    let mut flags: Vec<&str> = vec![
         "--target=wasm32",
-        irrt_cpp_path.to_str().unwrap(),
         "-x",
         "c++",
+        "-std=c++20",
         "-fno-discard-value-names",
         "-fno-exceptions",
         "-fno-rtti",
-        match env::var("PROFILE").as_deref() {
-            Ok("debug") => "-O0",
-            Ok("release") => "-O3",
-            flavor => panic!("Unknown or missing build flavor {flavor:?}"),
-        },
         "-emit-llvm",
         "-S",
         "-Wall",
         "-Wextra",
-        "-Werror=return-type",
-        "-I",
-        irrt_dir.to_str().unwrap(),
         "-o",
         "-",
+        "-I",
+        irrt_dir.to_str().unwrap(),
+        irrt_cpp_path.to_str().unwrap(),
     ];
 
-    println!("cargo:rerun-if-changed={}", out_dir.to_str().unwrap());
+    match env::var("PROFILE").as_deref() {
+        Ok("debug") => {
+            flags.push("-O0");
+            flags.push("-DIRRT_DEBUG_ASSERT");
+        }
+        Ok("release") => {
+            flags.push("-O3");
+        }
+        flavor => panic!("Unknown or missing build flavor {flavor:?}"),
+    }
 
+    // Tell Cargo to rerun if any file under `irrt_dir` (recursive) changes
+    println!("cargo:rerun-if-changed={}", irrt_dir.to_str().unwrap());
+
+    // Compile IRRT and capture the LLVM IR output
     let output = Command::new("clang-irrt")
         .args(flags)
         .output()
-        .map(|o| {
+        .inspect(|o| {
             assert!(o.status.success(), "{}", std::str::from_utf8(&o.stderr).unwrap());
-            o
         })
         .unwrap();
 
@@ -53,11 +65,17 @@ fn compile_irrt(irrt_dir: &Path, out_dir: &Path) {
     let output = std::str::from_utf8(&output.stdout).unwrap().replace("\r\n", "\n");
     let mut filtered_output = String::with_capacity(output.len());
 
-    // (?ms:^define.*?\}$) to capture `define` blocks
-    // (?m:^declare.*?$) to capture `declare` blocks
-    // (?m:^%.+?=\s*type\s*\{.+?\}$) to capture `type` declarations
-    let regex_filter =
-        Regex::new(r"(?ms:^define.*?\}$)|(?m:^declare.*?$)|(?m:^%.+?=\s*type\s*\{.+?\}$)").unwrap();
+    // Filter out irrelevant IR
+    //
+    // Regex:
+    // - `(?ms:^define.*?\}$)` captures LLVM `define` blocks
+    // - `(?m:^declare.*?$)` captures LLVM `declare` lines
+    // - `(?m:^%.+?=\s*type\s*\{.+?\}$)` captures LLVM `type` declarations
+    // - `(?m:^@.+?=.+$)` captures global constants
+    let regex_filter = Regex::new(
+        r"(?ms:^define.*?\}$)|(?m:^declare.*?$)|(?m:^%.+?=\s*type\s*\{.+?\}$)|(?m:^@.+?=.+$)",
+    )
+    .unwrap();
     for f in regex_filter.captures_iter(&output) {
         assert_eq!(f.len(), 1);
         filtered_output.push_str(&f[0]);
@@ -68,10 +86,14 @@ fn compile_irrt(irrt_dir: &Path, out_dir: &Path) {
         .unwrap()
         .replace_all(&filtered_output, "");
 
-    println!("cargo:rerun-if-env-changed=DEBUG_DUMP_IRRT");
-    if env::var("DEBUG_DUMP_IRRT").is_ok() {
+    // For debugging
+    // Doing `DEBUG_DUMP_IRRT=1 cargo build -p nac3core` dumps the LLVM IR generated
+    const DEBUG_DUMP_IRRT: &str = "DEBUG_DUMP_IRRT";
+    println!("cargo:rerun-if-env-changed={DEBUG_DUMP_IRRT}");
+    if env::var(DEBUG_DUMP_IRRT).is_ok() {
         let mut file = File::create(out_dir.join("irrt.ll")).unwrap();
         file.write_all(output.as_bytes()).unwrap();
+
         let mut file = File::create(out_dir.join("irrt-filtered.ll")).unwrap();
         file.write_all(filtered_output.as_bytes()).unwrap();
     }
@@ -85,50 +107,3 @@ fn compile_irrt(irrt_dir: &Path, out_dir: &Path) {
     llvm_as.stdin.as_mut().unwrap().write_all(filtered_output.as_bytes()).unwrap();
     assert!(llvm_as.wait().unwrap().success());
 }
-
-fn compile_irrt_test(irrt_dir: &Path, out_dir: &Path) {
-    let irrt_test_cpp_path = irrt_dir.join("irrt_test.cpp");
-    let exe_path = out_dir.join("irrt_test.out");
-
-    let flags: &[&str] = &[
-        irrt_test_cpp_path.to_str().unwrap(),
-        "-x",
-        "c++",
-        "-I",
-        irrt_dir.to_str().unwrap(),
-        "-g",
-        "-fno-discard-value-names",
-        "-O0",
-        "-Wall",
-        "-Wextra",
-        "-Werror=return-type",
-        "-lm", // for `tgamma()`, `lgamma()`
-        "-o",
-        exe_path.to_str().unwrap(),
-    ];
-
-    Command::new("clang-irrt-test")
-        .args(flags)
-        .output()
-        .map(|o| {
-            assert!(o.status.success(), "{}", std::str::from_utf8(&o.stderr).unwrap());
-            o
-        })
-        .unwrap();
-    println!("cargo:rerun-if-changed={}", out_dir.to_str().unwrap());
-}
-
-fn main() {
-    let out_dir = env::var("OUT_DIR").unwrap();
-    let out_dir = Path::new(&out_dir);
-
-    let irrt_dir = Path::new("./irrt");
-
-    compile_irrt(irrt_dir, out_dir);
-
-    // https://github.com/rust-lang/cargo/issues/2549
-    // `cargo test -F test` to also build `irrt_test.cpp
-    if cfg!(feature = "test") {
-        compile_irrt_test(irrt_dir, out_dir);
-    }
-}

nac3core/irrt/irrt.cpp

@@ -1,5 +1,5 @@
-#include "irrt_everything.hpp"
-
-/*
-    This file will be read by `clang-irrt` to conveniently produce LLVM IR for `nac3core/codegen`.
-*/
+#include "irrt/exception.hpp"
+#include "irrt/list.hpp"
+#include "irrt/math.hpp"
+#include "irrt/ndarray.hpp"
+#include "irrt/slice.hpp"

nac3core/irrt/irrt.hpp

@@ -1,437 +0,0 @@
-#ifndef IRRT_DONT_TYPEDEF_INTS
-typedef _BitInt(8) int8_t;
-typedef unsigned _BitInt(8) uint8_t;
-typedef _BitInt(32) int32_t;
-typedef unsigned _BitInt(32) uint32_t;
-typedef _BitInt(64) int64_t;
-typedef unsigned _BitInt(64) uint64_t;
-#endif
-
-// NDArray indices are always `uint32_t`.
-typedef uint32_t NDIndex;
-// The type of an index or a value describing the length of a range/slice is
-// always `int32_t`.
-typedef int32_t SliceIndex;
-
-template <typename T>
-static T max(T a, T b) {
-    return a > b ? a : b;
-}
-
-template <typename T>
-static T min(T a, T b) {
-    return a > b ? b : a;
-}
-
-// adapted from GNU Scientific Library: https://git.savannah.gnu.org/cgit/gsl.git/tree/sys/pow_int.c
-// need to make sure `exp >= 0` before calling this function
-template <typename T>
-static T __nac3_int_exp_impl(T base, T exp) {
-    T res = 1;
-    /* repeated squaring method */
-    do {
-        if (exp & 1) {
-            res *= base; /* for n odd */
-        }
-        exp >>= 1;
-        base *= base;
-    } while (exp);
-    return res;
-}
-
-template <typename SizeT>
-static SizeT __nac3_ndarray_calc_size_impl(
-    const SizeT *list_data,
-    SizeT list_len,
-    SizeT begin_idx,
-    SizeT end_idx
-) {
-    __builtin_assume(end_idx <= list_len);
-
-    SizeT num_elems = 1;
-    for (SizeT i = begin_idx; i < end_idx; ++i) {
-        SizeT val = list_data[i];
-        __builtin_assume(val > 0);
-        num_elems *= val;
-    }
-    return num_elems;
-}
-
-template <typename SizeT>
-static void __nac3_ndarray_calc_nd_indices_impl(
-    SizeT index,
-    const SizeT *dims,
-    SizeT num_dims,
-    NDIndex *idxs
-) {
-    SizeT stride = 1;
-    for (SizeT dim = 0; dim < num_dims; dim++) {
-        SizeT i = num_dims - dim - 1;
-        __builtin_assume(dims[i] > 0);
-        idxs[i] = (index / stride) % dims[i];
-        stride *= dims[i];
-    }
-}
-
-template <typename SizeT>
-static SizeT __nac3_ndarray_flatten_index_impl(
-    const SizeT *dims,
-    SizeT num_dims,
-    const NDIndex *indices,
-    SizeT num_indices
-) {
-    SizeT idx = 0;
-    SizeT stride = 1;
-    for (SizeT i = 0; i < num_dims; ++i) {
-        SizeT ri = num_dims - i - 1;
-        if (ri < num_indices) {
-            idx += stride * indices[ri];
-        }
-
-        __builtin_assume(dims[i] > 0);
-        stride *= dims[ri];
-    }
-    return idx;
-}
-
-template <typename SizeT>
-static void __nac3_ndarray_calc_broadcast_impl(
-    const SizeT *lhs_dims,
-    SizeT lhs_ndims,
-    const SizeT *rhs_dims,
-    SizeT rhs_ndims,
-    SizeT *out_dims
-) {
-    SizeT max_ndims = lhs_ndims > rhs_ndims ? lhs_ndims : rhs_ndims;
-
-    for (SizeT i = 0; i < max_ndims; ++i) {
-        const SizeT *lhs_dim_sz = i < lhs_ndims ? &lhs_dims[lhs_ndims - i - 1] : nullptr;
-        const SizeT *rhs_dim_sz = i < rhs_ndims ? &rhs_dims[rhs_ndims - i - 1] : nullptr;
-        SizeT *out_dim = &out_dims[max_ndims - i - 1];
-
-        if (lhs_dim_sz == nullptr) {
-            *out_dim = *rhs_dim_sz;
-        } else if (rhs_dim_sz == nullptr) {
-            *out_dim = *lhs_dim_sz;
-        } else if (*lhs_dim_sz == 1) {
-            *out_dim = *rhs_dim_sz;
-        } else if (*rhs_dim_sz == 1) {
-            *out_dim = *lhs_dim_sz;
-        } else if (*lhs_dim_sz == *rhs_dim_sz) {
-            *out_dim = *lhs_dim_sz;
-        } else {
-            __builtin_unreachable();
-        }
-    }
-}
-
-template <typename SizeT>
-static void __nac3_ndarray_calc_broadcast_idx_impl(
-    const SizeT *src_dims,
-    SizeT src_ndims,
-    const NDIndex *in_idx,
-    NDIndex *out_idx
-) {
-    for (SizeT i = 0; i < src_ndims; ++i) {
-        SizeT src_i = src_ndims - i - 1;
-        out_idx[src_i] = src_dims[src_i] == 1 ? 0 : in_idx[src_i];
-    }
-}
-
-template<typename SizeT>
-static void __nac3_ndarray_strides_from_shape_impl(
-    SizeT ndims,
-    SizeT *shape,
-    SizeT *dst_strides
-) {
-    SizeT stride_product = 1;
-    for (SizeT i = 0; i < ndims; i++) {
-        int dim_i = ndims - i - 1;
-        dst_strides[dim_i] = stride_product;
-        stride_product *= shape[dim_i];
-    }
-}
-
-extern "C" {
-    #define DEF_nac3_int_exp_(T) \
-        T __nac3_int_exp_##T(T base, T exp) {\
-            return __nac3_int_exp_impl(base, exp);\
-        }
-
-    DEF_nac3_int_exp_(int32_t)
-    DEF_nac3_int_exp_(int64_t)
-    DEF_nac3_int_exp_(uint32_t)
-    DEF_nac3_int_exp_(uint64_t)
-
-    SliceIndex __nac3_slice_index_bound(SliceIndex i, const SliceIndex len) {
-        if (i < 0) {
-            i = len + i;
-        }
-        if (i < 0) {
-            return 0;
-        } else if (i > len) {
-            return len;
-        }
-        return i;
-    }
-
-    SliceIndex __nac3_range_slice_len(
-        const SliceIndex start,
-        const SliceIndex end,
-        const SliceIndex step
-    ) {
-        SliceIndex diff = end - start;
-        if (diff > 0 && step > 0) {
-            return ((diff - 1) / step) + 1;
-        } else if (diff < 0 && step < 0) {
-            return ((diff + 1) / step) + 1;
-        } else {
-            return 0;
-        }
-    }
-
-    // Handle list assignment and dropping part of the list when
-    // both dest_step and src_step are +1.
-    // - All the index must *not* be out-of-bound or negative,
-    // - The end index is *inclusive*,
-    // - The length of src and dest slice size should already
-    // be checked: if dest.step == 1 then len(src) <= len(dest) else len(src) == len(dest)
-    SliceIndex __nac3_list_slice_assign_var_size(
-        SliceIndex dest_start,
-        SliceIndex dest_end,
-        SliceIndex dest_step,
-        uint8_t *dest_arr,
-        SliceIndex dest_arr_len,
-        SliceIndex src_start,
-        SliceIndex src_end,
-        SliceIndex src_step,
-        uint8_t *src_arr,
-        SliceIndex src_arr_len,
-        const SliceIndex size
-    ) {
-        /* if dest_arr_len == 0, do nothing since we do not support extending list */
-        if (dest_arr_len == 0) return dest_arr_len;
-        /* if both step is 1, memmove directly, handle the dropping of the list, and shrink size */
-        if (src_step == dest_step && dest_step == 1) {
-            const SliceIndex src_len = (src_end >= src_start) ? (src_end - src_start + 1) : 0;
-            const SliceIndex dest_len = (dest_end >= dest_start) ? (dest_end - dest_start + 1) : 0;
-            if (src_len > 0) {
-                __builtin_memmove(
-                    dest_arr + dest_start * size,
-                    src_arr + src_start * size,
-                    src_len * size
-                );
-            }
-            if (dest_len > 0) {
-                /* dropping */
-                __builtin_memmove(
-                    dest_arr + (dest_start + src_len) * size,
-                    dest_arr + (dest_end + 1) * size,
-                    (dest_arr_len - dest_end - 1) * size
-                );
-            }
-            /* shrink size */
-            return dest_arr_len - (dest_len - src_len);
-        }
-        /* if two range overlaps, need alloca */
-        uint8_t need_alloca =
-            (dest_arr == src_arr)
-            && !(
-                max(dest_start, dest_end) < min(src_start, src_end)
-                || max(src_start, src_end) < min(dest_start, dest_end)
-            );
-        if (need_alloca) {
-            uint8_t *tmp = reinterpret_cast<uint8_t *>(__builtin_alloca(src_arr_len * size));
-            __builtin_memcpy(tmp, src_arr, src_arr_len * size);
-            src_arr = tmp;
-        }
-        SliceIndex src_ind = src_start;
-        SliceIndex dest_ind = dest_start;
-        for (;
-            (src_step > 0) ? (src_ind <= src_end) : (src_ind >= src_end);
-            src_ind += src_step, dest_ind += dest_step
-        ) {
-            /* for constant optimization */
-            if (size == 1) {
-                __builtin_memcpy(dest_arr + dest_ind, src_arr + src_ind, 1);
-            } else if (size == 4) {
-                __builtin_memcpy(dest_arr + dest_ind * 4, src_arr + src_ind * 4, 4);
-            } else if (size == 8) {
-                __builtin_memcpy(dest_arr + dest_ind * 8, src_arr + src_ind * 8, 8);
-            } else {
-                /* memcpy for var size, cannot overlap after previous alloca */
-                __builtin_memcpy(dest_arr + dest_ind * size, src_arr + src_ind * size, size);
-            }
-        }
-        /* only dest_step == 1 can we shrink the dest list. */
-        /* size should be ensured prior to calling this function */
-        if (dest_step == 1 && dest_end >= dest_start) {
-            __builtin_memmove(
-                dest_arr + dest_ind * size,
-                dest_arr + (dest_end + 1) * size,
-                (dest_arr_len - dest_end - 1) * size
-            );
-            return dest_arr_len - (dest_end - dest_ind) - 1;
-        }
-        return dest_arr_len;
-    }
-
-    int32_t __nac3_isinf(double x) {
-        return __builtin_isinf(x);
-    }
-
-    int32_t __nac3_isnan(double x) {
-        return __builtin_isnan(x);
-    }
-
-    double tgamma(double arg);
-
-    double __nac3_gamma(double z) {
-        // Handling for denormals
-        //     | x                 | Python gamma(x) | C tgamma(x) |
-        // --- | ----------------- | --------------- | ----------- |
-        // (1) | nan               | nan             | nan         |
-        // (2) | -inf              | -inf            | inf         |
-        // (3) | inf               | inf             | inf         |
-        // (4) | 0.0               | inf             | inf         |
-        // (5) | {-1.0, -2.0, ...} | inf             | nan         |
-
-        // (1)-(3)
-        if (__builtin_isinf(z) || __builtin_isnan(z)) {
-            return z;
-        }
-
-        double v = tgamma(z);
-
-        // (4)-(5)
-        return __builtin_isinf(v) || __builtin_isnan(v) ? __builtin_inf() : v;
-    }
-
-    double lgamma(double arg);
-
-    double __nac3_gammaln(double x) {
-        // libm's handling of value overflows differs from scipy:
-        // - scipy: gammaln(-inf) -> -inf
-        // - libm : lgamma(-inf) -> inf
-
-        if (__builtin_isinf(x)) {
-            return x;
-        }
-
-        return lgamma(x);
-    }
-
-    double j0(double x);
-
-    double __nac3_j0(double x) {
-        // libm's handling of value overflows differs from scipy:
-        // - scipy: j0(inf) -> nan
-        // - libm : j0(inf) -> 0.0
-
-        if (__builtin_isinf(x)) {
-            return __builtin_nan("");
-        }
-
-        return j0(x);
-    }
-
-    uint32_t __nac3_ndarray_calc_size(
-        const uint32_t *list_data,
-        uint32_t list_len,
-        uint32_t begin_idx,
-        uint32_t end_idx
-    ) {
-        return __nac3_ndarray_calc_size_impl(list_data, list_len, begin_idx, end_idx);
-    }
-
-    uint64_t __nac3_ndarray_calc_size64(
-        const uint64_t *list_data,
-        uint64_t list_len,
-        uint64_t begin_idx,
-        uint64_t end_idx
-    ) {
-        return __nac3_ndarray_calc_size_impl(list_data, list_len, begin_idx, end_idx);
-    }
-
-    void __nac3_ndarray_calc_nd_indices(
-        uint32_t index,
-        const uint32_t* dims,
-        uint32_t num_dims,
-        NDIndex* idxs
-    ) {
-        __nac3_ndarray_calc_nd_indices_impl(index, dims, num_dims, idxs);
-    }
-
-    void __nac3_ndarray_calc_nd_indices64(
-        uint64_t index,
-        const uint64_t* dims,
-        uint64_t num_dims,
-        NDIndex* idxs
-    ) {
-        __nac3_ndarray_calc_nd_indices_impl(index, dims, num_dims, idxs);
-    }
-
-    uint32_t __nac3_ndarray_flatten_index(
-        const uint32_t* dims,
-        uint32_t num_dims,
-        const NDIndex* indices,
-        uint32_t num_indices
-    ) {
-        return __nac3_ndarray_flatten_index_impl(dims, num_dims, indices, num_indices);
-    }
-
-    uint64_t __nac3_ndarray_flatten_index64(
-        const uint64_t* dims,
-        uint64_t num_dims,
-        const NDIndex* indices,
-        uint64_t num_indices
-    ) {
-        return __nac3_ndarray_flatten_index_impl(dims, num_dims, indices, num_indices);
-    }
-
-    void __nac3_ndarray_calc_broadcast(
-        const uint32_t *lhs_dims,
-        uint32_t lhs_ndims,
-        const uint32_t *rhs_dims,
-        uint32_t rhs_ndims,
-        uint32_t *out_dims
-    ) {
-        return __nac3_ndarray_calc_broadcast_impl(lhs_dims, lhs_ndims, rhs_dims, rhs_ndims, out_dims);
-    }
-
-    void __nac3_ndarray_calc_broadcast64(
-        const uint64_t *lhs_dims,
-        uint64_t lhs_ndims,
-        const uint64_t *rhs_dims,
-        uint64_t rhs_ndims,
-        uint64_t *out_dims
-    ) {
-        return __nac3_ndarray_calc_broadcast_impl(lhs_dims, lhs_ndims, rhs_dims, rhs_ndims, out_dims);
-    }
-
-    void __nac3_ndarray_calc_broadcast_idx(
-        const uint32_t *src_dims,
-        uint32_t src_ndims,
-        const NDIndex *in_idx,
-        NDIndex *out_idx
-    ) {
-        __nac3_ndarray_calc_broadcast_idx_impl(src_dims, src_ndims, in_idx, out_idx);
-    }
-
-    void __nac3_ndarray_calc_broadcast_idx64(
-        const uint64_t *src_dims,
-        uint64_t src_ndims,
-        const NDIndex *in_idx,
-        NDIndex *out_idx
-    ) {
-        __nac3_ndarray_calc_broadcast_idx_impl(src_dims, src_ndims, in_idx, out_idx);
-    }
-
-    void __nac3_ndarray_strides_from_shape(uint32_t ndims, uint32_t* shape, uint32_t* dst_strides) {
-        __nac3_ndarray_strides_from_shape_impl(ndims, shape, dst_strides);
-    }
-
-    void __nac3_ndarray_strides_from_shape64(uint64_t ndims, uint64_t* shape, uint64_t* dst_strides) {
-        __nac3_ndarray_strides_from_shape_impl(ndims, shape, dst_strides);
-    }
-}

nac3core/irrt/irrt/cslice.hpp

@@ -0,0 +1,9 @@
+#pragma once
+
+#include "irrt/int_types.hpp"
+
+template<typename SizeT>
+struct CSlice {
+    void* base;
+    SizeT len;
+};

nac3core/irrt/irrt/debug.hpp

@@ -0,0 +1,25 @@
+#pragma once
+
+// Set in nac3core/build.rs
+#ifdef IRRT_DEBUG_ASSERT
+#define IRRT_DEBUG_ASSERT_BOOL true
+#else
+#define IRRT_DEBUG_ASSERT_BOOL false
+#endif
+
+#define raise_debug_assert(SizeT, msg, param1, param2, param3) \
+    raise_exception(SizeT, EXN_ASSERTION_ERROR, "IRRT debug assert failed: " msg, param1, param2, param3)
+
+#define debug_assert_eq(SizeT, lhs, rhs)                                           \
+    if constexpr (IRRT_DEBUG_ASSERT_BOOL) {                                        \
+        if ((lhs) != (rhs)) {                                                      \
+            raise_debug_assert(SizeT, "LHS = {0}. RHS = {1}", lhs, rhs, NO_PARAM); \
+        }                                                                          \
+    }
+
+#define debug_assert(SizeT, expr)                                                  \
+    if constexpr (IRRT_DEBUG_ASSERT_BOOL) {                                        \
+        if (!(expr)) {                                                             \
+            raise_debug_assert(SizeT, "Got false.", NO_PARAM, NO_PARAM, NO_PARAM); \
+        }                                                                          \
+    }

nac3core/irrt/irrt/exception.hpp

@@ -0,0 +1,85 @@
+#pragma once
+
+#include "irrt/cslice.hpp"
+#include "irrt/int_types.hpp"
+
+/**
+ * @brief The int type of ARTIQ exception IDs.
+ */
+using ExceptionId = int32_t;
+
+/*
+ * Set of exceptions C++ IRRT can use.
+ * Must be synchronized with `setup_irrt_exceptions` in `nac3core/src/codegen/irrt/mod.rs`.
+ */
+extern "C" {
+ExceptionId EXN_INDEX_ERROR;
+ExceptionId EXN_VALUE_ERROR;
+ExceptionId EXN_ASSERTION_ERROR;
+ExceptionId EXN_TYPE_ERROR;
+}
+
+/**
+ * @brief Extern function to `__nac3_raise`
+ *
+ * The parameter `err` could be `Exception<int32_t>` or `Exception<int64_t>`. The caller
+ * must make sure to pass `Exception`s with the correct `SizeT` depending on the `size_t` of the runtime.
+ */
+extern "C" void __nac3_raise(void* err);
+
+namespace {
+/**
+ * @brief NAC3's Exception struct
+ */
+template<typename SizeT>
+struct Exception {
+    ExceptionId id;
+    CSlice<SizeT> filename;
+    int32_t line;
+    int32_t column;
+    CSlice<SizeT> function;
+    CSlice<SizeT> msg;
+    int64_t params[3];
+};
+
+constexpr int64_t NO_PARAM = 0;
+
+template<typename SizeT>
+void _raise_exception_helper(ExceptionId id,
+                             const char* filename,
+                             int32_t line,
+                             const char* function,
+                             const char* msg,
+                             int64_t param0,
+                             int64_t param1,
+                             int64_t param2) {
+    Exception<SizeT> e = {
+        .id = id,
+        .filename = {.base = reinterpret_cast<void*>(const_cast<char*>(filename)),
+                     .len = static_cast<SizeT>(__builtin_strlen(filename))},
+        .line = line,
+        .column = 0,
+        .function = {.base = reinterpret_cast<void*>(const_cast<char*>(function)),
+                     .len = static_cast<SizeT>(__builtin_strlen(function))},
+        .msg = {.base = reinterpret_cast<void*>(const_cast<char*>(msg)),
+                .len = static_cast<SizeT>(__builtin_strlen(msg))},
+    };
+    e.params[0] = param0;
+    e.params[1] = param1;
+    e.params[2] = param2;
+    __nac3_raise(reinterpret_cast<void*>(&e));
+    __builtin_unreachable();
+}
+}  // namespace
+
+/**
+ * @brief Raise an exception with location details (location in the IRRT source files).
+ * @param SizeT The runtime `size_t` type.
+ * @param id The ID of the exception to raise.
+ * @param msg A global constant C-string of the error message.
+ *
+ * `param0` to `param2` are optional format arguments of `msg`. They should be set to
+ * `NO_PARAM` to indicate they are unused.
+ */
+#define raise_exception(SizeT, id, msg, param0, param1, param2) \
+    _raise_exception_helper<SizeT>(id, __FILE__, __LINE__, __FUNCTION__, msg, param0, param1, param2)

nac3core/irrt/irrt/int_types.hpp

@@ -0,0 +1,27 @@
+#pragma once
+
+#if __STDC_VERSION__ >= 202000
+using int8_t = _BitInt(8);
+using uint8_t = unsigned _BitInt(8);
+using int32_t = _BitInt(32);
+using uint32_t = unsigned _BitInt(32);
+using int64_t = _BitInt(64);
+using uint64_t = unsigned _BitInt(64);
+#else
+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Wdeprecated-type"
+using int8_t = _ExtInt(8);
+using uint8_t = unsigned _ExtInt(8);
+using int32_t = _ExtInt(32);
+using uint32_t = unsigned _ExtInt(32);
+using int64_t = _ExtInt(64);
+using uint64_t = unsigned _ExtInt(64);
+#pragma clang diagnostic pop
+
+#endif
+
+// NDArray indices are always `uint32_t`.
+using NDIndex = uint32_t;
+// The type of an index or a value describing the length of a range/slice is always `int32_t`.
+using SliceIndex = int32_t;

nac3core/irrt/irrt/list.hpp

@@ -0,0 +1,81 @@
+#pragma once
+
+#include "irrt/int_types.hpp"
+#include "irrt/math_util.hpp"
+
+extern "C" {
+// Handle list assignment and dropping part of the list when
+// both dest_step and src_step are +1.
+// - All the index must *not* be out-of-bound or negative,
+// - The end index is *inclusive*,
+// - The length of src and dest slice size should already
+// be checked: if dest.step == 1 then len(src) <= len(dest) else len(src) == len(dest)
+SliceIndex __nac3_list_slice_assign_var_size(SliceIndex dest_start,
+                                             SliceIndex dest_end,
+                                             SliceIndex dest_step,
+                                             void* dest_arr,
+                                             SliceIndex dest_arr_len,
+                                             SliceIndex src_start,
+                                             SliceIndex src_end,
+                                             SliceIndex src_step,
+                                             void* src_arr,
+                                             SliceIndex src_arr_len,
+                                             const SliceIndex size) {
+    /* if dest_arr_len == 0, do nothing since we do not support extending list */
+    if (dest_arr_len == 0)
+        return dest_arr_len;
+    /* if both step is 1, memmove directly, handle the dropping of the list, and shrink size */
+    if (src_step == dest_step && dest_step == 1) {
+        const SliceIndex src_len = (src_end >= src_start) ? (src_end - src_start + 1) : 0;
+        const SliceIndex dest_len = (dest_end >= dest_start) ? (dest_end - dest_start + 1) : 0;
+        if (src_len > 0) {
+            __builtin_memmove(static_cast<uint8_t*>(dest_arr) + dest_start * size,
+                              static_cast<uint8_t*>(src_arr) + src_start * size, src_len * size);
+        }
+        if (dest_len > 0) {
+            /* dropping */
+            __builtin_memmove(static_cast<uint8_t*>(dest_arr) + (dest_start + src_len) * size,
+                              static_cast<uint8_t*>(dest_arr) + (dest_end + 1) * size,
+                              (dest_arr_len - dest_end - 1) * size);
+        }
+        /* shrink size */
+        return dest_arr_len - (dest_len - src_len);
+    }
+    /* if two range overlaps, need alloca */
+    uint8_t need_alloca = (dest_arr == src_arr)
+                          && !(max(dest_start, dest_end) < min(src_start, src_end)
+                               || max(src_start, src_end) < min(dest_start, dest_end));
+    if (need_alloca) {
+        void* tmp = __builtin_alloca(src_arr_len * size);
+        __builtin_memcpy(tmp, src_arr, src_arr_len * size);
+        src_arr = tmp;
+    }
+    SliceIndex src_ind = src_start;
+    SliceIndex dest_ind = dest_start;
+    for (; (src_step > 0) ? (src_ind <= src_end) : (src_ind >= src_end); src_ind += src_step, dest_ind += dest_step) {
+        /* for constant optimization */
+        if (size == 1) {
+            __builtin_memcpy(static_cast<uint8_t*>(dest_arr) + dest_ind, static_cast<uint8_t*>(src_arr) + src_ind, 1);
+        } else if (size == 4) {
+            __builtin_memcpy(static_cast<uint8_t*>(dest_arr) + dest_ind * 4,
+                             static_cast<uint8_t*>(src_arr) + src_ind * 4, 4);
+        } else if (size == 8) {
+            __builtin_memcpy(static_cast<uint8_t*>(dest_arr) + dest_ind * 8,
+                             static_cast<uint8_t*>(src_arr) + src_ind * 8, 8);
+        } else {
+            /* memcpy for var size, cannot overlap after previous alloca */
+            __builtin_memcpy(static_cast<uint8_t*>(dest_arr) + dest_ind * size,
+                             static_cast<uint8_t*>(src_arr) + src_ind * size, size);
+        }
+    }
+    /* only dest_step == 1 can we shrink the dest list. */
+    /* size should be ensured prior to calling this function */
+    if (dest_step == 1 && dest_end >= dest_start) {
+        __builtin_memmove(static_cast<uint8_t*>(dest_arr) + dest_ind * size,
+                          static_cast<uint8_t*>(dest_arr) + (dest_end + 1) * size,
+                          (dest_arr_len - dest_end - 1) * size);
+        return dest_arr_len - (dest_end - dest_ind) - 1;
+    }
+    return dest_arr_len;
+}
+}  // extern "C"

nac3core/irrt/irrt/math.hpp

@@ -0,0 +1,93 @@
+#pragma once
+
+namespace {
+// adapted from GNU Scientific Library: https://git.savannah.gnu.org/cgit/gsl.git/tree/sys/pow_int.c
+// need to make sure `exp >= 0` before calling this function
+template<typename T>
+T __nac3_int_exp_impl(T base, T exp) {
+    T res = 1;
+    /* repeated squaring method */
+    do {
+        if (exp & 1) {
+            res *= base; /* for n odd */
+        }
+        exp >>= 1;
+        base *= base;
+    } while (exp);
+    return res;
+}
+}  // namespace
+
+#define DEF_nac3_int_exp_(T)                   \
+    T __nac3_int_exp_##T(T base, T exp) {      \
+        return __nac3_int_exp_impl(base, exp); \
+    }
+
+extern "C" {
+
+// Putting semicolons here to make clang-format not reformat this into
+// a stair shape.
+DEF_nac3_int_exp_(int32_t);
+DEF_nac3_int_exp_(int64_t);
+DEF_nac3_int_exp_(uint32_t);
+DEF_nac3_int_exp_(uint64_t);
+
+int32_t __nac3_isinf(double x) {
+    return __builtin_isinf(x);
+}
+
+int32_t __nac3_isnan(double x) {
+    return __builtin_isnan(x);
+}
+
+double tgamma(double arg);
+
+double __nac3_gamma(double z) {
+    // Handling for denormals
+    //     | x                 | Python gamma(x) | C tgamma(x) |
+    // --- | ----------------- | --------------- | ----------- |
+    // (1) | nan               | nan             | nan         |
+    // (2) | -inf              | -inf            | inf         |
+    // (3) | inf               | inf             | inf         |
+    // (4) | 0.0               | inf             | inf         |
+    // (5) | {-1.0, -2.0, ...} | inf             | nan         |
+
+    // (1)-(3)
+    if (__builtin_isinf(z) || __builtin_isnan(z)) {
+        return z;
+    }
+
+    double v = tgamma(z);
+
+    // (4)-(5)
+    return __builtin_isinf(v) || __builtin_isnan(v) ? __builtin_inf() : v;
+}
+
+double lgamma(double arg);
+
+double __nac3_gammaln(double x) {
+    // libm's handling of value overflows differs from scipy:
+    // - scipy: gammaln(-inf) -> -inf
+    // - libm : lgamma(-inf) -> inf
+
+    if (__builtin_isinf(x)) {
+        return x;
+    }
+
+    return lgamma(x);
+}
+
+double j0(double x);
+
+double __nac3_j0(double x) {
+    // libm's handling of value overflows differs from scipy:
+    // - scipy: j0(inf) -> nan
+    // - libm : j0(inf) -> 0.0
+
+    if (__builtin_isinf(x)) {
+        return __builtin_nan("");
+    }
+
+    return j0(x);
+}
+}  // namespace

nac3core/irrt/irrt/math_util.hpp

@@ -0,0 +1,13 @@
+#pragma once
+
+namespace {
+template<typename T>
+const T& max(const T& a, const T& b) {
+    return a > b ? a : b;
+}
+
+template<typename T>
+const T& min(const T& a, const T& b) {
+    return a > b ? b : a;
+}
+}  // namespace

nac3core/irrt/irrt/ndarray.hpp

@@ -0,0 +1,144 @@
+#pragma once
+
+#include "irrt/int_types.hpp"
+
+namespace {
+template<typename SizeT>
+SizeT __nac3_ndarray_calc_size_impl(const SizeT* list_data, SizeT list_len, SizeT begin_idx, SizeT end_idx) {
+    __builtin_assume(end_idx <= list_len);
+
+    SizeT num_elems = 1;
+    for (SizeT i = begin_idx; i < end_idx; ++i) {
+        SizeT val = list_data[i];
+        __builtin_assume(val > 0);
+        num_elems *= val;
+    }
+    return num_elems;
+}
+
+template<typename SizeT>
+void __nac3_ndarray_calc_nd_indices_impl(SizeT index, const SizeT* dims, SizeT num_dims, NDIndex* idxs) {
+    SizeT stride = 1;
+    for (SizeT dim = 0; dim < num_dims; dim++) {
+        SizeT i = num_dims - dim - 1;
+        __builtin_assume(dims[i] > 0);
+        idxs[i] = (index / stride) % dims[i];
+        stride *= dims[i];
+    }
+}
+
+template<typename SizeT>
+SizeT __nac3_ndarray_flatten_index_impl(const SizeT* dims, SizeT num_dims, const NDIndex* indices, SizeT num_indices) {
+    SizeT idx = 0;
+    SizeT stride = 1;
+    for (SizeT i = 0; i < num_dims; ++i) {
+        SizeT ri = num_dims - i - 1;
+        if (ri < num_indices) {
+            idx += stride * indices[ri];
+        }
+
+        __builtin_assume(dims[i] > 0);
+        stride *= dims[ri];
+    }
+    return idx;
+}
+
+template<typename SizeT>
+void __nac3_ndarray_calc_broadcast_impl(const SizeT* lhs_dims,
+                                        SizeT lhs_ndims,
+                                        const SizeT* rhs_dims,
+                                        SizeT rhs_ndims,
+                                        SizeT* out_dims) {
+    SizeT max_ndims = lhs_ndims > rhs_ndims ? lhs_ndims : rhs_ndims;
+
+    for (SizeT i = 0; i < max_ndims; ++i) {
+        const SizeT* lhs_dim_sz = i < lhs_ndims ? &lhs_dims[lhs_ndims - i - 1] : nullptr;
+        const SizeT* rhs_dim_sz = i < rhs_ndims ? &rhs_dims[rhs_ndims - i - 1] : nullptr;
+        SizeT* out_dim = &out_dims[max_ndims - i - 1];
+
+        if (lhs_dim_sz == nullptr) {
+            *out_dim = *rhs_dim_sz;
+        } else if (rhs_dim_sz == nullptr) {
+            *out_dim = *lhs_dim_sz;
+        } else if (*lhs_dim_sz == 1) {
+            *out_dim = *rhs_dim_sz;
+        } else if (*rhs_dim_sz == 1) {
+            *out_dim = *lhs_dim_sz;
+        } else if (*lhs_dim_sz == *rhs_dim_sz) {
+            *out_dim = *lhs_dim_sz;
+        } else {
+            __builtin_unreachable();
+        }
+    }
+}
+
+template<typename SizeT>
+void __nac3_ndarray_calc_broadcast_idx_impl(const SizeT* src_dims,
+                                            SizeT src_ndims,
+                                            const NDIndex* in_idx,
+                                            NDIndex* out_idx) {
+    for (SizeT i = 0; i < src_ndims; ++i) {
+        SizeT src_i = src_ndims - i - 1;
+        out_idx[src_i] = src_dims[src_i] == 1 ? 0 : in_idx[src_i];
+    }
+}
+}  // namespace
+
+extern "C" {
+uint32_t __nac3_ndarray_calc_size(const uint32_t* list_data, uint32_t list_len, uint32_t begin_idx, uint32_t end_idx) {
+    return __nac3_ndarray_calc_size_impl(list_data, list_len, begin_idx, end_idx);
+}
+
+uint64_t
+__nac3_ndarray_calc_size64(const uint64_t* list_data, uint64_t list_len, uint64_t begin_idx, uint64_t end_idx) {
+    return __nac3_ndarray_calc_size_impl(list_data, list_len, begin_idx, end_idx);
+}
+
+void __nac3_ndarray_calc_nd_indices(uint32_t index, const uint32_t* dims, uint32_t num_dims, NDIndex* idxs) {
+    __nac3_ndarray_calc_nd_indices_impl(index, dims, num_dims, idxs);
+}
+
+void __nac3_ndarray_calc_nd_indices64(uint64_t index, const uint64_t* dims, uint64_t num_dims, NDIndex* idxs) {
+    __nac3_ndarray_calc_nd_indices_impl(index, dims, num_dims, idxs);
+}
+
+uint32_t
+__nac3_ndarray_flatten_index(const uint32_t* dims, uint32_t num_dims, const NDIndex* indices, uint32_t num_indices) {
+    return __nac3_ndarray_flatten_index_impl(dims, num_dims, indices, num_indices);
+}
+
+uint64_t
+__nac3_ndarray_flatten_index64(const uint64_t* dims, uint64_t num_dims, const NDIndex* indices, uint64_t num_indices) {
+    return __nac3_ndarray_flatten_index_impl(dims, num_dims, indices, num_indices);
+}
+
+void __nac3_ndarray_calc_broadcast(const uint32_t* lhs_dims,
+                                   uint32_t lhs_ndims,
+                                   const uint32_t* rhs_dims,
+                                   uint32_t rhs_ndims,
+                                   uint32_t* out_dims) {
+    return __nac3_ndarray_calc_broadcast_impl(lhs_dims, lhs_ndims, rhs_dims, rhs_ndims, out_dims);
+}
+
+void __nac3_ndarray_calc_broadcast64(const uint64_t* lhs_dims,
+                                     uint64_t lhs_ndims,
+                                     const uint64_t* rhs_dims,
+                                     uint64_t rhs_ndims,
+                                     uint64_t* out_dims) {
+    return __nac3_ndarray_calc_broadcast_impl(lhs_dims, lhs_ndims, rhs_dims, rhs_ndims, out_dims);
+}
+
+void __nac3_ndarray_calc_broadcast_idx(const uint32_t* src_dims,
+                                       uint32_t src_ndims,
+                                       const NDIndex* in_idx,
+                                       NDIndex* out_idx) {
+    __nac3_ndarray_calc_broadcast_idx_impl(src_dims, src_ndims, in_idx, out_idx);
+}
+
+void __nac3_ndarray_calc_broadcast_idx64(const uint64_t* src_dims,
+                                         uint64_t src_ndims,
+                                         const NDIndex* in_idx,
+                                         NDIndex* out_idx) {
+    __nac3_ndarray_calc_broadcast_idx_impl(src_dims, src_ndims, in_idx, out_idx);
+}
+}  // namespace

nac3core/irrt/irrt/slice.hpp

@@ -0,0 +1,28 @@
+#pragma once
+
+#include "irrt/int_types.hpp"
+
+extern "C" {
+SliceIndex __nac3_slice_index_bound(SliceIndex i, const SliceIndex len) {
+    if (i < 0) {
+        i = len + i;
+    }
+    if (i < 0) {
+        return 0;
+    } else if (i > len) {
+        return len;
+    }
+    return i;
+}
+
+SliceIndex __nac3_range_slice_len(const SliceIndex start, const SliceIndex end, const SliceIndex step) {
+    SliceIndex diff = end - start;
+    if (diff > 0 && step > 0) {
+        return ((diff - 1) / step) + 1;
+    } else if (diff < 0 && step < 0) {
+        return ((diff + 1) / step) + 1;
+    } else {
+        return 0;
+    }
+}
+}  // namespace

nac3core/irrt/irrt_basic.hpp

@@ -1,216 +0,0 @@
-#pragma once
-
-#include "irrt_utils.hpp"
-#include "irrt_typedefs.hpp"
-
-/*
-    This header contains IRRT implementations
-    that do not deserved to be categorized (e.g., into numpy, etc.)
-
-    Check out other *.hpp files before including them here!!
-*/
-
-// The type of an index or a value describing the length of a range/slice is
-// always `int32_t`.
-
-namespace {
-    // adapted from GNU Scientific Library: https://git.savannah.gnu.org/cgit/gsl.git/tree/sys/pow_int.c
-    // need to make sure `exp >= 0` before calling this function
-    template <typename T>
-    T __nac3_int_exp_impl(T base, T exp) {
-        T res = 1;
-        /* repeated squaring method */
-        do {
-            if (exp & 1) {
-                res *= base; /* for n odd */
-            }
-            exp >>= 1;
-            base *= base;
-        } while (exp);
-        return res;
-    }
-}
-
-extern "C" {
-    #define DEF_nac3_int_exp_(T) \
-        T __nac3_int_exp_##T(T base, T exp) {\
-            return __nac3_int_exp_impl(base, exp);\
-        }
-
-    DEF_nac3_int_exp_(int32_t)
-    DEF_nac3_int_exp_(int64_t)
-    DEF_nac3_int_exp_(uint32_t)
-    DEF_nac3_int_exp_(uint64_t)
-
-    SliceIndex __nac3_slice_index_bound(SliceIndex i, const SliceIndex len) {
-        if (i < 0) {
-            i = len + i;
-        }
-        if (i < 0) {
-            return 0;
-        } else if (i > len) {
-            return len;
-        }
-        return i;
-    }
-
-    SliceIndex __nac3_range_slice_len(
-        const SliceIndex start,
-        const SliceIndex end,
-        const SliceIndex step
-    ) {
-        SliceIndex diff = end - start;
-        if (diff > 0 && step > 0) {
-            return ((diff - 1) / step) + 1;
-        } else if (diff < 0 && step < 0) {
-            return ((diff + 1) / step) + 1;
-        } else {
-            return 0;
-        }
-    }
-
-    // Handle list assignment and dropping part of the list when
-    // both dest_step and src_step are +1.
-    // - All the index must *not* be out-of-bound or negative,
-    // - The end index is *inclusive*,
-    // - The length of src and dest slice size should already
-    // be checked: if dest.step == 1 then len(src) <= len(dest) else len(src) == len(dest)
-    SliceIndex __nac3_list_slice_assign_var_size(
-        SliceIndex dest_start,
-        SliceIndex dest_end,
-        SliceIndex dest_step,
-        uint8_t *dest_arr,
-        SliceIndex dest_arr_len,
-        SliceIndex src_start,
-        SliceIndex src_end,
-        SliceIndex src_step,
-        uint8_t *src_arr,
-        SliceIndex src_arr_len,
-        const SliceIndex size
-    ) {
-        /* if dest_arr_len == 0, do nothing since we do not support extending list */
-        if (dest_arr_len == 0) return dest_arr_len;
-        /* if both step is 1, memmove directly, handle the dropping of the list, and shrink size */
-        if (src_step == dest_step && dest_step == 1) {
-            const SliceIndex src_len = (src_end >= src_start) ? (src_end - src_start + 1) : 0;
-            const SliceIndex dest_len = (dest_end >= dest_start) ? (dest_end - dest_start + 1) : 0;
-            if (src_len > 0) {
-                __builtin_memmove(
-                    dest_arr + dest_start * size,
-                    src_arr + src_start * size,
-                    src_len * size
-                );
-            }
-            if (dest_len > 0) {
-                /* dropping */
-                __builtin_memmove(
-                    dest_arr + (dest_start + src_len) * size,
-                    dest_arr + (dest_end + 1) * size,
-                    (dest_arr_len - dest_end - 1) * size
-                );
-            }
-            /* shrink size */
-            return dest_arr_len - (dest_len - src_len);
-        }
-        /* if two range overlaps, need alloca */
-        uint8_t need_alloca =
-            (dest_arr == src_arr)
-            && !(
-                max(dest_start, dest_end) < min(src_start, src_end)
-                || max(src_start, src_end) < min(dest_start, dest_end)
-            );
-        if (need_alloca) {
-            uint8_t *tmp = reinterpret_cast<uint8_t *>(__builtin_alloca(src_arr_len * size));
-            __builtin_memcpy(tmp, src_arr, src_arr_len * size);
-            src_arr = tmp;
-        }
-        SliceIndex src_ind = src_start;
-        SliceIndex dest_ind = dest_start;
-        for (;
-            (src_step > 0) ? (src_ind <= src_end) : (src_ind >= src_end);
-            src_ind += src_step, dest_ind += dest_step
-        ) {
-            /* for constant optimization */
-            if (size == 1) {
-                __builtin_memcpy(dest_arr + dest_ind, src_arr + src_ind, 1);
-            } else if (size == 4) {
-                __builtin_memcpy(dest_arr + dest_ind * 4, src_arr + src_ind * 4, 4);
-            } else if (size == 8) {
-                __builtin_memcpy(dest_arr + dest_ind * 8, src_arr + src_ind * 8, 8);
-            } else {
-                /* memcpy for var size, cannot overlap after previous alloca */
-                __builtin_memcpy(dest_arr + dest_ind * size, src_arr + src_ind * size, size);
-            }
-        }
-        /* only dest_step == 1 can we shrink the dest list. */
-        /* size should be ensured prior to calling this function */
-        if (dest_step == 1 && dest_end >= dest_start) {
-            __builtin_memmove(
-                dest_arr + dest_ind * size,
-                dest_arr + (dest_end + 1) * size,
-                (dest_arr_len - dest_end - 1) * size
-            );
-            return dest_arr_len - (dest_end - dest_ind) - 1;
-        }
-        return dest_arr_len;
-    }
-
-    int32_t __nac3_isinf(double x) {
-        return __builtin_isinf(x);
-    }
-
-    int32_t __nac3_isnan(double x) {
-        return __builtin_isnan(x);
-    }
-
-    double tgamma(double arg);
-
-    double __nac3_gamma(double z) {
-        // Handling for denormals
-        //     | x                 | Python gamma(x) | C tgamma(x) |
-        // --- | ----------------- | --------------- | ----------- |
-        // (1) | nan               | nan             | nan         |
-        // (2) | -inf              | -inf            | inf         |
-        // (3) | inf               | inf             | inf         |
-        // (4) | 0.0               | inf             | inf         |
-        // (5) | {-1.0, -2.0, ...} | inf             | nan         |
-
-        // (1)-(3)
-        if (__builtin_isinf(z) || __builtin_isnan(z)) {
-            return z;
-        }
-
-        double v = tgamma(z);
-
-        // (4)-(5)
-        return __builtin_isinf(v) || __builtin_isnan(v) ? __builtin_inf() : v;
-    }
-
-    double lgamma(double arg);
-
-    double __nac3_gammaln(double x) {
-        // libm's handling of value overflows differs from scipy:
-        // - scipy: gammaln(-inf) -> -inf
-        // - libm : lgamma(-inf) -> inf
-
-        if (__builtin_isinf(x)) {
-            return x;
-        }
-
-        return lgamma(x);
-    }
-
-    double j0(double x);
-
-    double __nac3_j0(double x) {
-        // libm's handling of value overflows differs from scipy:
-        // - scipy: j0(inf) -> nan
-        // - libm : j0(inf) -> 0.0
-
-        if (__builtin_isinf(x)) {
-            return __builtin_nan("");
-        }
-
-        return j0(x);
-    }
-}

nac3core/irrt/irrt_everything.hpp

@@ -1,14 +0,0 @@
-#pragma once
-
-#include "irrt_utils.hpp"
-#include "irrt_typedefs.hpp"
-#include "irrt_basic.hpp"
-#include "irrt_slice.hpp"
-#include "irrt_numpy_ndarray.hpp"
-
-/*
-    All IRRT implementations.
-
-    We don't have any pre-compiled objects, so we are writing all implementations in headers and
-    concatenate them with `#include` into one massive source file that contains all the IRRT stuff.
-*/

nac3core/irrt/irrt_numpy_ndarray.hpp

@@ -1,466 +0,0 @@
-#pragma once
-
-#include "irrt_utils.hpp"
-#include "irrt_typedefs.hpp"
-#include "irrt_slice.hpp"
-
-/*
-    NDArray-related implementations.
-`*/
-
-// NDArray indices are always `uint32_t`.
-using NDIndex = uint32_t;
-
-namespace {
-    namespace ndarray_util {
-        template <typename SizeT>
-        static void set_indices_by_nth(SizeT ndims, const SizeT* shape, SizeT* indices, SizeT nth) {
-            for (int32_t i = 0; i < ndims; i++) {
-                int32_t dim_i = ndims - i - 1;
-                int32_t dim = shape[dim_i];
-
-                indices[dim_i] = nth % dim;
-                nth /= dim;
-            }
-        }
-
-        // Compute the strides of an ndarray given an ndarray `shape`
-        // and assuming that the ndarray is *fully C-contagious*.
-        //
-        // You might want to read up on https://ajcr.net/stride-guide-part-1/.
-        template <typename SizeT>
-        static void set_strides_by_shape(SizeT itemsize, SizeT ndims, SizeT* dst_strides, const SizeT* shape) {
-            SizeT stride_product = 1;
-            for (SizeT i = 0; i < ndims; i++) {
-                int dim_i = ndims - i - 1;
-                dst_strides[dim_i] = stride_product * itemsize;
-                stride_product *= shape[dim_i];
-            }
-        }
-
-        // Compute the size/# of elements of an ndarray given its shape
-        template <typename SizeT>
-        static SizeT calc_size_from_shape(SizeT ndims, const SizeT* shape) {
-            SizeT size = 1;
-            for (SizeT dim_i = 0; dim_i < ndims; dim_i++) size *= shape[dim_i];
-            return size;
-        }
-
-        template <typename SizeT>
-        static bool can_broadcast_shape_to(
-            const SizeT target_ndims,
-            const SizeT *target_shape,
-            const SizeT src_ndims,
-            const SizeT *src_shape
-        ) {
-            /*
-                // See https://numpy.org/doc/stable/user/basics.broadcasting.html
-
-                This function handles this example:
-                ```
-                Image  (3d array): 256 x 256 x 3
-                Scale  (1d array):             3
-                Result (3d array): 256 x 256 x 3
-                ```
-
-                Other interesting examples to consider:
-                - `can_broadcast_shape_to([3], [1, 1, 1, 1, 3]) == true`
-                - `can_broadcast_shape_to([3], [3, 1]) == false`
-                - `can_broadcast_shape_to([256, 256, 3], [256, 1, 3]) == true`
-
-                In cases when the shapes contain zero(es):
-                - `can_broadcast_shape_to([0], [1]) == true`
-                - `can_broadcast_shape_to([0], [2]) == false`
-                - `can_broadcast_shape_to([0, 4, 0, 0], [1]) == true`
-                - `can_broadcast_shape_to([0, 4, 0, 0], [1, 1, 1, 1]) == true`
-                - `can_broadcast_shape_to([0, 4, 0, 0], [1, 4, 1, 1]) == true`
-                - `can_broadcast_shape_to([4, 3], [0, 3]) == false`
-                - `can_broadcast_shape_to([4, 3], [0, 0]) == false`
-            */
-
-            // This is essentially doing the following in Python:
-            // `for target_dim, src_dim in itertools.zip_longest(target_shape[::-1], src_shape[::-1], fillvalue=1)`
-            for (SizeT i = 0; i < max(target_ndims, src_ndims); i++) {
-                SizeT target_dim_i = target_ndims - i - 1;
-                SizeT src_dim_i = src_ndims - i - 1;
-
-                bool target_dim_exists = target_dim_i >= 0;
-                bool src_dim_exists = src_dim_i >= 0;
-
-                SizeT target_dim = target_dim_exists ? target_shape[target_dim_i] : 1;
-                SizeT src_dim = src_dim_exists ? src_shape[src_dim_i] : 1;
-
-                bool ok = src_dim == 1 || target_dim == src_dim;
-                if (!ok) return false;
-            }
-
-            return true;
-        }
-    }
-
-    typedef uint8_t NDSliceType;
-    extern "C" {
-        const NDSliceType INPUT_SLICE_TYPE_INDEX = 0;
-        const NDSliceType INPUT_SLICE_TYPE_SLICE = 1;
-    }
-
-    struct NDSlice {
-        // A poor-man's `std::variant<int, UserRange>`
-        NDSliceType type;
-
-        /*
-            if type == INPUT_SLICE_TYPE_INDEX => `slice` points to a single `SizeT`
-            if type == INPUT_SLICE_TYPE_SLICE => `slice` points to a single `UserRange`
-        */
-        uint8_t *slice;
-    };
-
-    namespace ndarray_util {
-        template<typename SizeT>
-        SizeT deduce_ndims_after_slicing(SizeT ndims, SizeT num_slices, const NDSlice *slices) {
-            irrt_assert(num_slices <= ndims);
-
-            SizeT final_ndims = ndims;
-            for (SizeT i = 0; i < num_slices; i++) {
-                if (slices[i].type == INPUT_SLICE_TYPE_INDEX) {
-                    final_ndims--; // An integer slice demotes the rank by 1
-                }
-            }
-            return final_ndims;
-        }
-    }
-
-    template <typename SizeT>
-    struct NDArrayIndicesIter {
-        SizeT ndims;
-        const SizeT *shape;
-        SizeT *indices;
-
-        void set_indices_zero() {
-            __builtin_memset(indices, 0, sizeof(SizeT) * ndims);
-        }
-
-        void next() {
-            for (SizeT i = 0; i < ndims; i++) {
-                SizeT dim_i = ndims - i - 1;
-
-                indices[dim_i]++;
-                if (indices[dim_i] < shape[dim_i]) {
-                    break;
-                } else {
-                    indices[dim_i] = 0;
-                }
-            }
-        }
-    };
-
-    // The NDArray object. `SizeT` is the *signed* size type of this ndarray.
-    //
-    // NOTE: The order of fields is IMPORTANT. DON'T TOUCH IT
-    //
-    // Some resources you might find helpful:
-    // - The official numpy implementations:
-    //   - https://github.com/numpy/numpy/blob/735a477f0bc2b5b84d0e72d92f224bde78d4e069/doc/source/reference/c-api/types-and-structures.rst
-    // - On strides (about reshaping, slicing, C-contagiousness, etc)
-    //   - https://ajcr.net/stride-guide-part-1/.
-    //   - https://ajcr.net/stride-guide-part-2/.
-    //   - https://ajcr.net/stride-guide-part-3/.
-    template <typename SizeT>
-    struct NDArray {
-        // The underlying data this `ndarray` is pointing to.
-        //
-        // NOTE: Formally this should be of type `void *`, but clang
-        // translates `void *` to `i8 *` when run with `-S -emit-llvm`,
-        // so we will put `uint8_t *` here for clarity.
-        uint8_t *data;
-
-        // The number of bytes of a single element in `data`.
-        //
-        // The `SizeT` is treated as `unsigned`.
-        SizeT itemsize;
-
-        // The number of dimensions of this shape.
-        //
-        // The `SizeT` is treated as `unsigned`.
-        SizeT ndims;
-
-        // Array shape, with length equal to `ndims`.
-        //
-        // The `SizeT` is treated as `unsigned`.
-        //
-        // NOTE: `shape` can contain 0.
-        //   (those appear when the user makes an out of bounds slice into an ndarray, e.g., `np.zeros((3, 3))[400:].shape == (0, 3)`)
-        SizeT *shape;
-
-        // Array strides (stride value is in number of bytes, NOT number of elements), with length equal to `ndims`.
-        //
-        // The `SizeT` is treated as `signed`.
-        //
-        // NOTE: `strides` can have negative numbers.
-        //   (those appear when there is a slice with a negative step, e.g., `my_array[::-1]`)
-        SizeT *strides;
-
-        // Calculate the size/# of elements of an `ndarray`.
-        // This function corresponds to `np.size(<ndarray>)` or `ndarray.size`
-        SizeT size() {
-            return ndarray_util::calc_size_from_shape(ndims, shape);
-        }
-
-        // Calculate the number of bytes of its content of an `ndarray` *in its view*.
-        // This function corresponds to `ndarray.nbytes`
-        SizeT nbytes() {
-            return this->size() * itemsize;
-        }
-
-        void set_value_at_pelement(uint8_t* pelement, const uint8_t* pvalue) {
-            __builtin_memcpy(pelement, pvalue, itemsize);
-        }
-
-        uint8_t* get_pelement(const SizeT *indices) {
-            uint8_t* element = data;
-            for (SizeT dim_i = 0; dim_i < ndims; dim_i++)
-                element += indices[dim_i] * strides[dim_i];
-            return element;
-        }
-
-        uint8_t* get_nth_pelement(SizeT nth) {
-            irrt_assert(0 <= nth);
-            irrt_assert(nth < this->size());
-
-            SizeT* indices = (SizeT*) __builtin_alloca(sizeof(SizeT) * this->ndims);
-            ndarray_util::set_indices_by_nth(this->ndims, this->shape, indices, nth);
-            return get_pelement(indices);
-        }
-
-        // Get pointer to the first element of this ndarray, assuming
-        // `this->size() > 0`, i.e., not "degenerate" due to zeroes in `this->shape`)
-        //
-        // This is particularly useful for when the ndarray is just containing a single scalar.
-        uint8_t* get_first_pelement() {
-            irrt_assert(this->size() > 0);
-            return this->data; // ...It is simply `this->data`
-        }
-
-        // Is the given `indices` valid/in-bounds?
-        bool in_bounds(const SizeT *indices) {
-            for (SizeT dim_i = 0; dim_i < ndims; dim_i++) {
-                bool dim_ok = indices[dim_i] < shape[dim_i];
-                if (!dim_ok) return false;
-            }
-            return true;
-        }
-
-        // Fill the ndarray with a value
-        void fill_generic(const uint8_t* pvalue) {
-            NDArrayIndicesIter<SizeT> iter;
-            iter.ndims = this->ndims;
-            iter.shape = this->shape;
-            iter.indices = (SizeT*) __builtin_alloca(sizeof(SizeT) * ndims);
-            iter.set_indices_zero();
-
-            for (SizeT i = 0; i < this->size(); i++, iter.next()) {
-                uint8_t* pelement = get_pelement(iter.indices);
-                set_value_at_pelement(pelement, pvalue);
-            }
-        }
-
-        // Set the strides of the ndarray with `ndarray_util::set_strides_by_shape`
-        void set_strides_by_shape() {
-            ndarray_util::set_strides_by_shape(itemsize, ndims, strides, shape);
-        }
-
-        // https://numpy.org/doc/stable/reference/generated/numpy.eye.html
-        void set_to_eye(SizeT k, const uint8_t* zero_pvalue, const uint8_t* one_pvalue) {
-            __builtin_assume(ndims == 2);
-
-            // TODO: Better implementation
-
-            fill_generic(zero_pvalue);
-            for (SizeT i = 0; i < min(shape[0], shape[1]); i++) {
-                SizeT row = i;
-                SizeT col = i + k;
-                SizeT indices[2] = { row, col };
-
-                if (!in_bounds(indices)) continue;
-
-                uint8_t* pelement = get_pelement(indices);
-                set_value_at_pelement(pelement, one_pvalue);
-            }
-        }
-
-        // To support numpy complex slices (e.g., `my_array[:50:2,4,:2:-1]`)
-        //
-        // Things assumed by this function:
-        // - `dst_ndarray` is allocated by the caller
-        // - `dst_ndarray.ndims` has the correct value (according to `ndarray_util::deduce_ndims_after_slicing`).
-        //   - ... and `dst_ndarray.shape` and `dst_ndarray.strides` have been allocated by the caller as well
-        //
-        // Other notes:
-        // - `dst_ndarray->data` does not have to be set, it will be derived.
-        // - `dst_ndarray->itemsize` does not have to be set, it will be set to `this->itemsize`
-        // - `dst_ndarray->shape` and `dst_ndarray.strides` can contain empty values
-        void slice(SizeT num_ndslices, NDSlice* ndslices, NDArray<SizeT>* dst_ndarray) {
-            // REFERENCE CODE (check out `_index_helper` in `__getitem__`):
-            //   https://github.com/wadetb/tinynumpy/blob/0d23d22e07062ffab2afa287374c7b366eebdda1/tinynumpy/tinynumpy.py#L652
-
-            irrt_assert(dst_ndarray->ndims == ndarray_util::deduce_ndims_after_slicing(this->ndims, num_ndslices, ndslices));
-
-            dst_ndarray->data = this->data;
-
-            SizeT this_axis = 0;
-            SizeT dst_axis = 0;
-
-            for (SizeT i = 0; i < num_ndslices; i++) {
-                NDSlice *ndslice = &ndslices[i];
-                if (ndslice->type == INPUT_SLICE_TYPE_INDEX) {
-                    // Handle when the ndslice is just a single (possibly negative) integer
-                    // e.g., `my_array[::2, -5, ::-1]`
-                    //                      ^^------ like this
-                    SizeT index_user = *((SizeT*) ndslice->slice);
-                    SizeT index = resolve_index_in_length(this->shape[this_axis], index_user);
-                    dst_ndarray->data += index * this->strides[this_axis]; // Add offset
-
-                    // Next
-                    this_axis++;
-                } else if (ndslice->type == INPUT_SLICE_TYPE_SLICE) {
-                    // Handle when the ndslice is a slice (represented by UserSlice in IRRT)
-                    // e.g., `my_array[::2, -5, ::-1]`
-                    //                 ^^^------^^^^----- like these
-                    UserSlice<SizeT>* user_slice = (UserSlice<SizeT>*) ndslice->slice;
-                    Slice<SizeT> slice = user_slice->indices(this->shape[this_axis]); // To resolve negative indices and other funny stuff written by the user
-
-                    // NOTE: There is no need to write special code to handle negative steps/strides.
-                    // This simple implementation meticulously handles both positive and negative steps/strides.
-                    // Check out the tinynumpy and IRRT's test cases if you are not convinced.
-                    dst_ndarray->data += slice.start * this->strides[this_axis]; // Add offset (NOTE: no need to `* itemsize`, strides count in # of bytes)
-                    dst_ndarray->strides[dst_axis] = slice.step * this->strides[this_axis]; // Determine stride
-                    dst_ndarray->shape[dst_axis] = slice.len(); // Determine shape dimension
-
-                    // Next
-                    dst_axis++;
-                    this_axis++;
-                } else {
-                    __builtin_unreachable();
-                }
-            }
-
-            irrt_assert(dst_axis == dst_ndarray->ndims); // Sanity check on the implementation
-        }
-
-        // Similar to `np.broadcast_to(<ndarray>, <target_shape>)`
-        // Assumptions:
-        //   - `this` has to be fully initialized.
-        //   - `dst_ndarray->ndims` has to be set.
-        //   - `dst_ndarray->shape` has to be set, this determines the shape `this` broadcasts to.
-        //
-        // Other notes:
-        //   - `dst_ndarray->data` does not have to be set, it will be set to `this->data`.
-        //   - `dst_ndarray->itemsize` does not have to be set, it will be set to `this->data`.
-        //   - `dst_ndarray->strides` does not have to be set, it will be overwritten.
-        //
-        // Cautions:
-        //   ```
-        //   xs = np.zeros((4,))
-        //   ys = np.zero((4, 1))
-        //   ys[:] = xs # ok
-        // 
-        //   xs = np.zeros((1, 4))
-        //   ys = np.zero((4,))
-        //   ys[:] = xs # allowed
-        //   # However `np.broadcast_to(xs, (4,))` would fails, as per numpy's broadcasting rule.
-        //   # and apparently numpy will "deprecate" this? SEE https://github.com/numpy/numpy/issues/21744
-        //   # This implementation will NOT support this assignment.
-        //   ```
-        void broadcast_to(NDArray<SizeT>* dst_ndarray) {
-            dst_ndarray->data = this->data;
-            dst_ndarray->itemsize = this->itemsize;
-
-            irrt_assert(
-                ndarray_util::can_broadcast_shape_to(
-                    dst_ndarray->ndims,
-                    dst_ndarray->shape,
-                    this->ndims,
-                    this->shape
-                )
-            );
-
-            SizeT stride_product = 1;
-            for (SizeT i = 0; i < max(this->ndims, dst_ndarray->ndims); i++) {
-                SizeT this_dim_i = this->ndims - i - 1;
-                SizeT dst_dim_i = dst_ndarray->ndims - i - 1;
-
-                bool this_dim_exists = this_dim_i >= 0;
-                bool dst_dim_exists = dst_dim_i >= 0;
-
-                // TODO: Explain how this works
-                bool c1 = this_dim_exists && this->shape[this_dim_i] == 1;
-                bool c2 = dst_dim_exists && dst_ndarray->shape[dst_dim_i] != 1;
-                if (!this_dim_exists || (c1 && c2)) {
-                    dst_ndarray->strides[dst_dim_i] = 0; // Freeze it in-place
-                } else {
-                    dst_ndarray->strides[dst_dim_i] = stride_product * this->itemsize;
-                    stride_product *= this->shape[this_dim_i]; // NOTE: this_dim_exist must be true here.
-                }
-            }
-        }
-
-        // Simulates `this_ndarray[:] = src_ndarray`, with automatic broadcasting.
-        // Caution on https://github.com/numpy/numpy/issues/21744
-        // Also see `NDArray::broadcast_to`
-        void assign_with(NDArray<SizeT>* src_ndarray) {
-            irrt_assert(
-                ndarray_util::can_broadcast_shape_to(
-                    this->ndims,
-                    this->shape,
-                    src_ndarray->ndims,
-                    src_ndarray->shape
-                )
-            );
-
-            // Broadcast the `src_ndarray` to make the reading process *much* easier
-            SizeT* broadcasted_src_ndarray_strides = __builtin_alloca(sizeof(SizeT) * this->ndims); // Remember to allocate strides beforehand
-            NDArray<SizeT> broadcasted_src_ndarray = {
-                .ndims = this->ndims,
-                .shape = this->shape,
-                .strides = broadcasted_src_ndarray_strides
-            };
-            src_ndarray->broadcast_to(&broadcasted_src_ndarray);
-
-            // Using iter instead of `get_nth_pelement` because it is slightly faster
-            SizeT* indices = __builtin_alloca(sizeof(SizeT) * this->ndims);
-            auto iter = NDArrayIndicesIter<SizeT> {
-                .ndims = this->ndims,
-                .shape = this->shape,
-                .indices = indices
-            };
-            const SizeT this_size = this->size();
-            for (SizeT i = 0; i < this_size; i++, iter.next()) {
-                uint8_t* src_pelement = broadcasted_src_ndarray_strides->get_pelement(indices);
-                uint8_t* this_pelement = this->get_pelement(indices);
-                this->set_value_at_pelement(src_pelement, src_pelement);
-            }
-        }
-    };
-}
-
-extern "C" {
-    uint32_t __nac3_ndarray_size(NDArray<int32_t>* ndarray) {
-        return ndarray->size();
-    }
-
-    uint64_t __nac3_ndarray_size64(NDArray<int64_t>* ndarray) {
-        return ndarray->size();
-    }
-
-    void __nac3_ndarray_fill_generic(NDArray<int32_t>* ndarray, uint8_t* pvalue) {
-        ndarray->fill_generic(pvalue);
-    }
-
-    void __nac3_ndarray_fill_generic64(NDArray<int64_t>* ndarray, uint8_t* pvalue) {
-        ndarray->fill_generic(pvalue);
-    }
-
-    // void __nac3_ndarray_slice(NDArray<int32_t>* ndarray, int32_t num_slices, NDSlice<int32_t> *slices, NDArray<int32_t> *dst_ndarray) {
-    //     // ndarray->slice(num_slices, slices, dst_ndarray);
-    // }
-}

nac3core/irrt/irrt_slice.hpp

@@ -1,80 +0,0 @@
-#pragma once
-
-#include "irrt_utils.hpp"
-#include "irrt_typedefs.hpp"
-
-namespace {
-    // A proper slice in IRRT, all negative indices have be resolved to absolute values.
-    // Even though nac3core's slices are always `int32_t`, we will template slice anyway
-    // since this struct is used as a general utility.
-    template <typename T>
-    struct Slice {
-        T start;
-        T stop;
-        T step;
-
-        // The length/The number of elements of the slice if it were a range,
-        // i.e., the value of `len(range(this->start, this->stop, this->end))`
-        T len() {
-            T diff = stop - start;
-            if (diff > 0 && step > 0) {
-                return ((diff - 1) / step) + 1;
-            } else if (diff < 0 && step < 0) {
-                return ((diff + 1) / step) + 1;
-            } else {
-                return 0;
-            }
-        }
-    };
-
-    template<typename T>
-    T resolve_index_in_length(T length, T index) {
-        irrt_assert(length >= 0);
-        if (index < 0) {
-            // Remember that index is negative, so do a plus here
-            return max(length + index, 0);
-        } else {
-            return min(length, index);
-        }
-    }
-
-    // NOTE: using a bitfield for the `*_defined` is better, at the
-    // cost of a more annoying implementation in nac3core inkwell
-    template <typename T>
-    struct UserSlice {
-        uint8_t start_defined;
-        T start;
-
-        uint8_t stop_defined;
-        T stop;
-
-        uint8_t step_defined;
-        T step;
-
-        // Like Python's `slice(start, stop, step).indices(length)`
-        Slice<T> indices(T length) {
-            // NOTE: This function implements Python's `slice.indices` *FAITHFULLY*.
-            // SEE: https://github.com/python/cpython/blob/f62161837e68c1c77961435f1b954412dd5c2b65/Objects/sliceobject.c#L546
-            irrt_assert(length >= 0);
-            irrt_assert(!step_defined || step != 0); // step_defined -> step != 0; step cannot be zero if specified by user
-
-            Slice<T> result;
-            result.step = step_defined ? step : 1;
-            bool step_is_negative = result.step < 0;
-
-            if (start_defined) {
-                result.start = resolve_index_in_length(length, start);
-            } else {
-                result.start = step_is_negative ? length - 1 : 0;
-            }
-            
-            if (stop_defined) {
-                result.stop = resolve_index_in_length(length, stop);
-            } else {
-                result.stop = step_is_negative ? -1 : length;
-            }
-
-            return result;
-        }
-    };
-}

nac3core/irrt/irrt_test.cpp

@@ -1,658 +0,0 @@
-// This file will be compiled like a real C++ program,
-// and we do have the luxury to use the standard libraries.
-// That is if the nix flakes do not have issues... especially on msys2...
-#include <cstdint>
-#include <cstdio>
-#include <cstdlib>
-
-// Set `IRRT_DONT_TYPEDEF_INTS` because `cstdint` defines them
-#define IRRT_DONT_TYPEDEF_INTS
-#include "irrt_everything.hpp"
-
-void test_fail() {
-    printf("[!] Test failed\n");
-    exit(1);
-}
-
-void __begin_test(const char* function_name, const char* file, int line) {
-    printf("######### Running %s @ %s:%d\n", function_name, file, line);
-}
-
-#define BEGIN_TEST() __begin_test(__FUNCTION__, __FILE__, __LINE__)
-
-template <typename T>
-void debug_print_array(const char* format, int len, T* as) {
-    printf("[");
-    for (int i = 0; i < len; i++) {
-        if (i != 0) printf(", ");
-        printf(format, as[i]);
-    }
-    printf("]");
-}
-
-template <typename T>
-void assert_arrays_match(const char* label, const char* format, int len, T* expected, T* got) {
-    if (!arrays_match(len, expected, got)) {
-        printf(">>>>>>> %s\n", label);
-        printf("    Expecting = ");
-        debug_print_array(format, len, expected);
-        printf("\n");
-        printf("          Got = ");
-        debug_print_array(format, len, got);
-        printf("\n");
-        test_fail();
-    }
-}
-
-template <typename T>
-void assert_values_match(const char* label, const char* format, T expected, T got) {
-    if (expected != got) {
-        printf(">>>>>>> %s\n", label);
-        printf("    Expecting = ");
-        printf(format, expected);
-        printf("\n");
-        printf("          Got = ");
-        printf(format, got);
-        printf("\n");
-        test_fail();
-    }
-}
-
-void print_repeated(const char *str, int count) {
-    for (int i = 0; i < count; i++) {
-        printf("%s", str);
-    }
-}
-
-template<typename SizeT, typename ElementT>
-void __print_ndarray_aux(const char *format, bool first, bool last, SizeT* cursor, SizeT depth, NDArray<SizeT>* ndarray) {
-    // A really lazy recursive implementation
-    
-    // Add left padding unless its the first entry (since there would be "[[[" before it)
-    if (!first) {
-        print_repeated(" ", depth);
-    }
-
-    const SizeT dim = ndarray->shape[depth];
-    if (depth + 1 == ndarray->ndims) {
-        // Recursed down to last dimension, print the values in a nice list
-        printf("[");
-
-        SizeT* indices = (SizeT*) __builtin_alloca(sizeof(SizeT) * ndarray->ndims);
-        for (SizeT i = 0; i < dim; i++) {
-            ndarray_util::set_indices_by_nth(ndarray->ndims, ndarray->shape, indices, *cursor);
-            ElementT* pelement = (ElementT*) ndarray->get_pelement(indices);
-            ElementT element = *pelement;
-
-            if (i != 0) printf(", "); // List delimiter
-            printf(format, element);
-            printf("(@");
-            debug_print_array("%d", ndarray->ndims, indices);
-            printf(")");
-
-            (*cursor)++;
-        }
-        printf("]");
-    } else {
-        printf("[");
-        for (SizeT i = 0; i < ndarray->shape[depth]; i++) {
-            __print_ndarray_aux<SizeT, ElementT>(
-                format,
-                i == 0, // first?
-                i + 1 == dim, // last?
-                cursor,
-                depth + 1,
-                ndarray
-            );
-        }
-        printf("]");
-    }
-
-    // Add newline unless its the last entry (since there will be "]]]" after it)
-    if (!last) {
-        print_repeated("\n", depth);
-    }
-}
-
-template<typename SizeT, typename ElementT>
-void print_ndarray(const char *format, NDArray<SizeT>* ndarray) {
-    if (ndarray->ndims == 0) {
-        printf("<empty ndarray>");
-    } else {
-        SizeT cursor = 0;
-        __print_ndarray_aux<SizeT, ElementT>(format, true, true, &cursor, 0, ndarray);
-    }
-    printf("\n");
-}
-
-void test_calc_size_from_shape_normal() {
-    // Test shapes with normal values
-    BEGIN_TEST();
-
-    int32_t shape[4] = { 2, 3, 5, 7 };
-    assert_values_match("size", "%d", 210, ndarray_util::calc_size_from_shape<int32_t>(4, shape));
-}
-
-void test_calc_size_from_shape_has_zero() {
-    // Test shapes with 0 in them
-    BEGIN_TEST();
-
-    int32_t shape[4] = { 2, 0, 5, 7 };
-    assert_values_match("size", "%d", 0, ndarray_util::calc_size_from_shape<int32_t>(4, shape));
-}
-
-void test_set_strides_by_shape() {
-    // Test `set_strides_by_shape()`
-    BEGIN_TEST();
-
-    int32_t shape[4] = { 99, 3, 5, 7 };
-    int32_t strides[4] = { 0 };
-    ndarray_util::set_strides_by_shape((int32_t) sizeof(int32_t), 4, strides, shape);
-
-    int32_t expected_strides[4] = {
-        105 * sizeof(int32_t),
-        35 * sizeof(int32_t),
-        7 * sizeof(int32_t),
-        1 * sizeof(int32_t)
-    };
-    assert_arrays_match("strides", "%u", 4u, expected_strides, strides);
-}
-
-void test_ndarray_indices_iter_normal() {
-    // Test NDArrayIndicesIter normal behavior
-    BEGIN_TEST();
-
-    int32_t shape[3] = { 1, 2, 3 };
-    int32_t indices[3] = { 0, 0, 0 };
-    auto iter = NDArrayIndicesIter<int32_t> {
-        .ndims = 3,
-        .shape = shape,
-        .indices = indices
-    };
-
-    assert_arrays_match("indices #0", "%u", 3u, iter.indices, (int32_t[3]) { 0, 0, 0 });
-    iter.next();
-    assert_arrays_match("indices #1", "%u", 3u, iter.indices, (int32_t[3]) { 0, 0, 1 });
-    iter.next();
-    assert_arrays_match("indices #2", "%u", 3u, iter.indices, (int32_t[3]) { 0, 0, 2 });
-    iter.next();
-    assert_arrays_match("indices #3", "%u", 3u, iter.indices, (int32_t[3]) { 0, 1, 0 });
-    iter.next();
-    assert_arrays_match("indices #4", "%u", 3u, iter.indices, (int32_t[3]) { 0, 1, 1 });
-    iter.next();
-    assert_arrays_match("indices #5", "%u", 3u, iter.indices, (int32_t[3]) { 0, 1, 2 });
-    iter.next();
-    assert_arrays_match("indices #6", "%u", 3u, iter.indices, (int32_t[3]) { 0, 0, 0 }); // Loops back
-    iter.next();
-    assert_arrays_match("indices #7", "%u", 3u, iter.indices, (int32_t[3]) { 0, 0, 1 });
-}
-
-void test_ndarray_fill_generic() {
-    // Test ndarray fill_generic
-    BEGIN_TEST();
-
-    // Choose a type that's neither int32_t nor uint64_t (candidates of SizeT) to spice it up
-    // Also make all the octets non-zero, to see if `memcpy` in `fill_generic` is working perfectly.
-    uint16_t fill_value = 0xFACE;
-
-    uint16_t in_data[6] = { 100, 101, 102, 103, 104, 105 }; // Fill `data` with values that != `999`
-    int32_t in_itemsize = sizeof(uint16_t);
-    const int32_t in_ndims = 2;
-    int32_t in_shape[in_ndims] = { 2, 3 };
-    int32_t in_strides[in_ndims] = {};
-    NDArray<int32_t> ndarray = {
-        .data = (uint8_t*) in_data,
-        .itemsize = in_itemsize,
-        .ndims = in_ndims,
-        .shape = in_shape,
-        .strides = in_strides,
-    };
-    ndarray.set_strides_by_shape();
-    ndarray.fill_generic((uint8_t*) &fill_value); // `fill_generic` here
-
-    uint16_t expected_data[6] = { fill_value, fill_value, fill_value, fill_value, fill_value, fill_value };
-    assert_arrays_match("data", "0x%hX", 6, expected_data, in_data);
-}
-
-void test_ndarray_set_to_eye() {
-    // Test `set_to_eye` behavior (helper function to implement `np.eye()`)
-    BEGIN_TEST();
-
-    double in_data[9] = { 99.0, 99.0, 99.0, 99.0, 99.0, 99.0, 99.0, 99.0, 99.0 };
-    int32_t in_itemsize = sizeof(double);
-    const int32_t in_ndims = 2;
-    int32_t in_shape[in_ndims] = { 3, 3 };
-    int32_t in_strides[in_ndims] = {};
-    NDArray<int32_t> ndarray = {
-        .data = (uint8_t*) in_data,
-        .itemsize = in_itemsize,
-        .ndims = in_ndims,
-        .shape = in_shape,
-        .strides = in_strides,
-    };
-    ndarray.set_strides_by_shape();
-
-    double zero = 0.0;
-    double one = 1.0;
-    ndarray.set_to_eye(1, (uint8_t*) &zero, (uint8_t*) &one);
-
-    assert_values_match("in_data[0]", "%f", 0.0, in_data[0]);
-    assert_values_match("in_data[1]", "%f", 1.0, in_data[1]);
-    assert_values_match("in_data[2]", "%f", 0.0, in_data[2]);
-    assert_values_match("in_data[3]", "%f", 0.0, in_data[3]);
-    assert_values_match("in_data[4]", "%f", 0.0, in_data[4]);
-    assert_values_match("in_data[5]", "%f", 1.0, in_data[5]);
-    assert_values_match("in_data[6]", "%f", 0.0, in_data[6]);
-    assert_values_match("in_data[7]", "%f", 0.0, in_data[7]);
-    assert_values_match("in_data[8]", "%f", 0.0, in_data[8]);
-}
-
-void test_slice_1() {
-    // Test `slice(5, None, None).indices(100) == slice(5, 100, 1)`
-    BEGIN_TEST();
-
-    UserSlice<int> user_slice = {
-        .start_defined = 1,
-        .start = 5,
-        .stop_defined = 0,
-        .step_defined = 0,
-    };
-
-    auto slice = user_slice.indices(100);
-    assert_values_match("start", "%d", 5, slice.start);
-    assert_values_match("stop", "%d", 100, slice.stop);
-    assert_values_match("step", "%d", 1, slice.step);
-}
-
-void test_slice_2() {
-    // Test `slice(400, 999, None).indices(100) == slice(100, 100, 1)`
-    BEGIN_TEST();
-
-    UserSlice<int> user_slice = {
-        .start_defined = 1,
-        .start = 400,
-        .stop_defined = 0,
-        .step_defined = 0,
-    };
-
-    auto slice = user_slice.indices(100);
-    assert_values_match("start", "%d", 100, slice.start);
-    assert_values_match("stop", "%d", 100, slice.stop);
-    assert_values_match("step", "%d", 1, slice.step);
-}
-
-void test_slice_3() {
-    // Test `slice(-10, -5, None).indices(100) == slice(90, 95, 1)`
-    BEGIN_TEST();
-
-    UserSlice<int> user_slice = {
-        .start_defined = 1,
-        .start = -10,
-        .stop_defined = 1,
-        .stop = -5,
-        .step_defined = 0,
-    };
-
-    auto slice = user_slice.indices(100);
-    assert_values_match("start", "%d", 90, slice.start);
-    assert_values_match("stop", "%d", 95, slice.stop);
-    assert_values_match("step", "%d", 1, slice.step);
-}
-
-void test_slice_4() {
-    // Test `slice(None, None, -5).indices(100) == (99, -1, -5)`
-    BEGIN_TEST();
-
-    UserSlice<int> user_slice = {
-        .start_defined = 0,
-        .stop_defined = 0,
-        .step_defined = 1,
-        .step = -5
-    };
-
-    auto slice = user_slice.indices(100);
-    assert_values_match("start", "%d", 99, slice.start);
-    assert_values_match("stop", "%d", -1, slice.stop);
-    assert_values_match("step", "%d", -5, slice.step);
-}
-
-void test_ndslice_1() {
-    /*
-        Reference Python code:
-        ```python
-        ndarray = np.arange(12, dtype=np.float64).reshape((3, 4));
-        # array([[ 0.,  1.,  2.,  3.],
-        #        [ 4.,  5.,  6.,  7.],
-        #        [ 8.,  9., 10., 11.]])
-
-        dst_ndarray = ndarray[-2:, 1::2]
-        # array([[ 5.,  7.],
-        #        [ 9., 11.]])
-
-        assert dst_ndarray.shape == (2, 2)
-        assert dst_ndarray.strides == (32, 16)
-        assert dst_ndarray[0, 0] == 5.0
-        assert dst_ndarray[0, 1] == 7.0
-        assert dst_ndarray[1, 0] == 9.0
-        assert dst_ndarray[1, 1] == 11.0
-        ```
-    */
-    BEGIN_TEST();
-
-    double in_data[12] = { 0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0 };
-    int32_t in_itemsize = sizeof(double);
-    const int32_t in_ndims = 2;
-    int32_t in_shape[in_ndims] = { 3, 4 };
-    int32_t in_strides[in_ndims] = {};
-    NDArray<int32_t> ndarray = {
-        .data = (uint8_t*) in_data,
-        .itemsize = in_itemsize,
-        .ndims = in_ndims,
-        .shape = in_shape,
-        .strides = in_strides
-    };
-    ndarray.set_strides_by_shape();
-
-    // Destination ndarray
-    // As documented, ndims and shape & strides must be allocated and determined by the caller.
-    const int32_t dst_ndims = 2;
-    int32_t dst_shape[dst_ndims] = {999, 999}; // Empty values
-    int32_t dst_strides[dst_ndims] = {999, 999}; // Empty values
-    NDArray<int32_t> dst_ndarray = {
-        .data = nullptr,
-        .ndims = dst_ndims,
-        .shape = dst_shape,
-        .strides = dst_strides
-    };
-
-    // Create the slice in `ndarray[-2::, 1::2]`
-    UserSlice<int32_t> user_slice_1 = {
-        .start_defined = 1,
-        .start = -2,
-        .stop_defined = 0,
-        .step_defined = 0
-    };
-
-    UserSlice<int32_t> user_slice_2 = {
-        .start_defined = 1,
-        .start = 1,
-        .stop_defined = 0,
-        .step_defined = 1,
-        .step = 2
-    };
-
-    const int32_t num_ndslices = 2;
-    NDSlice ndslices[num_ndslices] = {
-        { .type = INPUT_SLICE_TYPE_SLICE, .slice = (uint8_t*) &user_slice_1 },
-        { .type = INPUT_SLICE_TYPE_SLICE, .slice = (uint8_t*) &user_slice_2 }
-    };
-
-    ndarray.slice(num_ndslices, ndslices, &dst_ndarray);
-
-    int32_t expected_shape[dst_ndims] = { 2, 2 };
-    int32_t expected_strides[dst_ndims] = { 32, 16 };
-    assert_arrays_match("shape", "%d", dst_ndims, expected_shape, dst_ndarray.shape);
-    assert_arrays_match("strides", "%d", dst_ndims, expected_strides, dst_ndarray.strides);
-
-    assert_values_match("dst_ndarray[0, 0]", "%f", 5.0, *((double *) dst_ndarray.get_pelement((int32_t[dst_ndims]) { 0, 0 })));
-    assert_values_match("dst_ndarray[0, 1]", "%f", 7.0, *((double *) dst_ndarray.get_pelement((int32_t[dst_ndims]) { 0, 1 })));
-    assert_values_match("dst_ndarray[1, 0]", "%f", 9.0, *((double *) dst_ndarray.get_pelement((int32_t[dst_ndims]) { 1, 0 })));
-    assert_values_match("dst_ndarray[1, 1]", "%f", 11.0, *((double *) dst_ndarray.get_pelement((int32_t[dst_ndims]) { 1, 1 })));
-}
-
-void test_ndslice_2() {
-    /*
-        ```python
-        ndarray = np.arange(12, dtype=np.float64).reshape((3, 4))
-        # array([[ 0.,  1.,  2.,  3.],
-        #        [ 4.,  5.,  6.,  7.],
-        #        [ 8.,  9., 10., 11.]])
-
-        dst_ndarray = ndarray[2, ::-2]
-        # array([11.,  9.])
-
-        assert dst_ndarray.shape == (2,)
-        assert dst_ndarray.strides == (-16,)
-        assert dst_ndarray[0] == 11.0
-        assert dst_ndarray[1] == 9.0
-
-        dst_ndarray[1, 0] == 99 # If you write to `dst_ndarray`
-        assert ndarray[1, 3] == 99 # `ndarray` also updates!!
-        ```
-    */
-    BEGIN_TEST();
-
-    double in_data[12] = { 0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0 };
-    int32_t in_itemsize = sizeof(double);
-    const int32_t in_ndims = 2;
-    int32_t in_shape[in_ndims] = { 3, 4 };
-    int32_t in_strides[in_ndims] = {};
-    NDArray<int32_t> ndarray = {
-        .data = (uint8_t*) in_data,
-        .itemsize = in_itemsize,
-        .ndims = in_ndims,
-        .shape = in_shape,
-        .strides = in_strides
-    };
-    ndarray.set_strides_by_shape();
-
-    // Destination ndarray
-    // As documented, ndims and shape & strides must be allocated and determined by the caller.
-    const int32_t dst_ndims = 1;
-    int32_t dst_shape[dst_ndims] = {999}; // Empty values
-    int32_t dst_strides[dst_ndims] = {999}; // Empty values
-    NDArray<int32_t> dst_ndarray = {
-        .data = nullptr,
-        .ndims = dst_ndims,
-        .shape = dst_shape,
-        .strides = dst_strides
-    };
-
-    // Create the slice in `ndarray[2, ::-2]`
-    int32_t user_slice_1 = 2;
-    UserSlice<int32_t> user_slice_2 = {
-        .start_defined = 0,
-        .stop_defined = 0,
-        .step_defined = 1,
-        .step = -2
-    };
-
-    const int32_t num_ndslices = 2;
-    NDSlice ndslices[num_ndslices] = {
-        { .type = INPUT_SLICE_TYPE_INDEX, .slice = (uint8_t*) &user_slice_1 },
-        { .type = INPUT_SLICE_TYPE_SLICE, .slice = (uint8_t*) &user_slice_2 }
-    };
-
-    ndarray.slice(num_ndslices, ndslices, &dst_ndarray);
-
-    int32_t expected_shape[dst_ndims] = { 2 };
-    int32_t expected_strides[dst_ndims] = { -16 };
-    assert_arrays_match("shape", "%d", dst_ndims, expected_shape, dst_ndarray.shape);
-    assert_arrays_match("strides", "%d", dst_ndims, expected_strides, dst_ndarray.strides);
-
-    // [5.0, 3.0]
-    assert_values_match("dst_ndarray[0]", "%f", 11.0, *((double *) dst_ndarray.get_pelement((int32_t[dst_ndims]) { 0 })));
-    assert_values_match("dst_ndarray[1]", "%f", 9.0, *((double *) dst_ndarray.get_pelement((int32_t[dst_ndims]) { 1 })));
-}
-
-void test_can_broadcast_shape() {
-    BEGIN_TEST();
-
-    assert_values_match(
-        "can_broadcast_shape_to([3], [1, 1, 1, 1, 3]) == true",
-        "%d",
-        true,
-        ndarray_util::can_broadcast_shape_to(1, (int32_t[]) { 3 }, 5, (int32_t[]) { 1, 1, 1, 1, 3 })
-    );
-    assert_values_match(
-        "can_broadcast_shape_to([3], [3, 1]) == false",
-        "%d",
-        false,
-        ndarray_util::can_broadcast_shape_to(1, (int32_t[]) { 3 }, 2, (int32_t[]) { 3, 1 }));
-    assert_values_match(
-        "can_broadcast_shape_to([3], [3]) == true",
-        "%d",
-        true,
-        ndarray_util::can_broadcast_shape_to(1, (int32_t[]) { 3 }, 1, (int32_t[]) { 3 }));
-    assert_values_match(
-        "can_broadcast_shape_to([1], [3]) == false",
-        "%d",
-        false,
-        ndarray_util::can_broadcast_shape_to(1, (int32_t[]) { 1 }, 1, (int32_t[]) { 3 }));
-    assert_values_match(
-        "can_broadcast_shape_to([1], [1]) == true",
-        "%d",
-        true,
-        ndarray_util::can_broadcast_shape_to(1, (int32_t[]) { 1 }, 1, (int32_t[]) { 1 }));
-    assert_values_match(
-        "can_broadcast_shape_to([256, 256, 3], [256, 1, 3]) == true",
-        "%d",
-        true,
-        ndarray_util::can_broadcast_shape_to(3, (int32_t[]) { 256, 256, 3 }, 3, (int32_t[]) { 256, 1, 3 })
-    );
-    assert_values_match(
-        "can_broadcast_shape_to([256, 256, 3], [3]) == true",
-        "%d",
-        true,
-        ndarray_util::can_broadcast_shape_to(3, (int32_t[]) { 256, 256, 3 }, 1, (int32_t[]) { 3 })
-    );
-    assert_values_match(
-        "can_broadcast_shape_to([256, 256, 3], [2]) == false",
-        "%d",
-        false,
-        ndarray_util::can_broadcast_shape_to(3, (int32_t[]) { 256, 256, 3 }, 1, (int32_t[]) { 2 })
-    );
-    assert_values_match(
-        "can_broadcast_shape_to([256, 256, 3], [1]) == true",
-        "%d",
-        true,
-        ndarray_util::can_broadcast_shape_to(3, (int32_t[]) { 256, 256, 3 }, 1, (int32_t[]) { 1 })
-    );
-
-    // In cases when the shapes contain zero(es)
-    assert_values_match(
-        "can_broadcast_shape_to([0], [1]) == true",
-        "%d",
-        true,
-        ndarray_util::can_broadcast_shape_to(1, (int32_t[]) { 0 }, 1, (int32_t[]) { 1 })
-    );
-    assert_values_match(
-        "can_broadcast_shape_to([0], [2]) == false",
-        "%d",
-        false,
-        ndarray_util::can_broadcast_shape_to(1, (int32_t[]) { 0 }, 1, (int32_t[]) { 2 })
-    );
-    assert_values_match(
-        "can_broadcast_shape_to([0, 4, 0, 0], [1]) == true",
-        "%d",
-        true,
-        ndarray_util::can_broadcast_shape_to(4, (int32_t[]) { 0, 4, 0, 0 }, 1, (int32_t[]) { 1 })
-    );
-    assert_values_match(
-        "can_broadcast_shape_to([0, 4, 0, 0], [1, 1, 1, 1]) == true",
-        "%d",
-        true,
-        ndarray_util::can_broadcast_shape_to(4, (int32_t[]) { 0, 4, 0, 0 }, 4, (int32_t[]) { 1, 1, 1, 1 })
-    );
-    assert_values_match(
-        "can_broadcast_shape_to([0, 4, 0, 0], [1, 4, 1, 1]) == true",
-        "%d",
-        true,
-        ndarray_util::can_broadcast_shape_to(4, (int32_t[]) { 0, 4, 0, 0 }, 4, (int32_t[]) { 1, 4, 1, 1 })
-    );
-    assert_values_match(
-        "can_broadcast_shape_to([4, 3], [0, 3]) == false",
-        "%d",
-        false,
-        ndarray_util::can_broadcast_shape_to(2, (int32_t[]) { 4, 3 }, 2, (int32_t[]) { 0, 3 })
-    );
-    assert_values_match(
-        "can_broadcast_shape_to([4, 3], [0, 0]) == false",
-        "%d",
-        false,
-        ndarray_util::can_broadcast_shape_to(2, (int32_t[]) { 4, 3 }, 2, (int32_t[]) { 0, 0 })
-    );
-}
-
-void test_ndarray_broadcast_1() {
-    /*
-    # array = np.array([[19.9, 29.9, 39.9, 49.9]], dtype=np.float64)
-    # >>> [[19.9 29.9 39.9 49.9]]
-    #
-    # array = np.broadcast_to(array, (2, 3, 4))
-    # >>> [[[19.9 29.9 39.9 49.9]
-    # >>>   [19.9 29.9 39.9 49.9]
-    # >>>   [19.9 29.9 39.9 49.9]]
-    # >>>  [[19.9 29.9 39.9 49.9]
-    # >>>   [19.9 29.9 39.9 49.9]
-    # >>>   [19.9 29.9 39.9 49.9]]]
-    #
-    # assery array.strides == (0, 0, 8)
-
-    */
-    BEGIN_TEST();
-
-    double in_data[4] = { 19.9, 29.9, 39.9, 49.9 };
-    const int32_t in_ndims = 2;
-    int32_t in_shape[in_ndims] = {1, 4};
-    int32_t in_strides[in_ndims] = {};
-    NDArray<int32_t> ndarray = {
-        .data = (uint8_t*) in_data,
-        .itemsize = sizeof(double),
-        .ndims = in_ndims,
-        .shape = in_shape,
-        .strides = in_strides
-    };
-    ndarray.set_strides_by_shape();
-
-    const int32_t dst_ndims = 3;
-    int32_t dst_shape[dst_ndims] = {2, 3, 4};
-    int32_t dst_strides[dst_ndims] = {};
-    NDArray<int32_t> dst_ndarray = {
-        .ndims = dst_ndims,
-        .shape = dst_shape,
-        .strides = dst_strides
-    };
-
-    ndarray.broadcast_to(&dst_ndarray);
-
-    assert_arrays_match("dst_ndarray->strides", "%d", dst_ndims, (int32_t[]) { 0, 0, 8 }, dst_ndarray.strides);
-
-    assert_values_match("dst_ndarray[0, 0, 0]", "%f", 19.9, *((double*) dst_ndarray.get_pelement((int32_t[]) {0, 0, 0})));
-    assert_values_match("dst_ndarray[0, 0, 1]", "%f", 29.9, *((double*) dst_ndarray.get_pelement((int32_t[]) {0, 0, 1})));
-    assert_values_match("dst_ndarray[0, 0, 2]", "%f", 39.9, *((double*) dst_ndarray.get_pelement((int32_t[]) {0, 0, 2})));
-    assert_values_match("dst_ndarray[0, 0, 3]", "%f", 49.9, *((double*) dst_ndarray.get_pelement((int32_t[]) {0, 0, 3})));
-    assert_values_match("dst_ndarray[0, 1, 0]", "%f", 19.9, *((double*) dst_ndarray.get_pelement((int32_t[]) {0, 1, 0})));
-    assert_values_match("dst_ndarray[0, 1, 1]", "%f", 29.9, *((double*) dst_ndarray.get_pelement((int32_t[]) {0, 1, 1})));
-    assert_values_match("dst_ndarray[0, 1, 2]", "%f", 39.9, *((double*) dst_ndarray.get_pelement((int32_t[]) {0, 1, 2})));
-    assert_values_match("dst_ndarray[0, 1, 3]", "%f", 49.9, *((double*) dst_ndarray.get_pelement((int32_t[]) {0, 1, 3})));
-    assert_values_match("dst_ndarray[1, 2, 3]", "%f", 49.9, *((double*) dst_ndarray.get_pelement((int32_t[]) {1, 2, 3})));
-}
-
-void test_assign_with() {
-    /*
-        ```
-        xs = np.array([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0], [7.0, 8.0, 9.0]], dtype=np.float64)
-        ys = xs.shape
-        ```
-    */
-}
-
-int main() {
-    test_calc_size_from_shape_normal();
-    test_calc_size_from_shape_has_zero();
-    test_set_strides_by_shape();
-    test_ndarray_indices_iter_normal();
-    test_ndarray_fill_generic();
-    test_ndarray_set_to_eye();
-    test_slice_1();
-    test_slice_2();
-    test_slice_3();
-    test_slice_4();
-    test_ndslice_1();
-    test_ndslice_2();
-    test_can_broadcast_shape();
-    test_ndarray_broadcast_1();
-    test_assign_with();
-    return 0;
-}

nac3core/irrt/irrt_typedefs.hpp

@@ -1,14 +0,0 @@
-#pragma once
-
-// This is made toggleable since `irrt_test.cpp` itself would include
-// headers that define the `int_t` family.
-#ifndef IRRT_DONT_TYPEDEF_INTS
-typedef _BitInt(8) int8_t;
-typedef unsigned _BitInt(8) uint8_t;
-typedef _BitInt(32) int32_t;
-typedef unsigned _BitInt(32) uint32_t;
-typedef _BitInt(64) int64_t;
-typedef unsigned _BitInt(64) uint64_t;
-#endif
-
-typedef int32_t SliceIndex;

nac3core/irrt/irrt_utils.hpp

@@ -1,37 +0,0 @@
-#pragma once
-
-#include "irrt_typedefs.hpp"
-
-namespace {
-    template <typename T>
-    T max(T a, T b) {
-        return a > b ? a : b;
-    }
-
-    template <typename T>
-    T min(T a, T b) {
-        return a > b ? b : a;
-    }
-
-    template <typename T>
-    bool arrays_match(int len, T *as, T *bs) {
-        for (int i = 0; i < len; i++) {
-            if (as[i] != bs[i]) return false;
-        }
-        return true;
-    }
-
-    void irrt_panic() {
-        // Crash the program for now.
-        // TODO: Don't crash the program
-        //       ... or at least produce a good message when doing testing IRRT
-            
-        uint8_t* death = nullptr;
-        *death = 0; // TODO: address 0 on hardware might be writable?
-    }
-
-    // TODO: Make this a macro and allow it to be toggled on/off (e.g., debug vs release)
-    void irrt_assert(bool condition) {
-        if (!condition) irrt_panic();
-    }
-}

nac3core/nac3core_derive/Cargo.toml

@@ -0,0 +1,21 @@
+[package]
+name = "nac3core_derive"
+version = "0.1.0"
+edition = "2021"
+
+[lib]
+proc-macro = true
+
+[[test]]
+name = "structfields_tests"
+path = "tests/structfields_test.rs"
+
+[dev-dependencies]
+nac3core = { path = ".." }
+trybuild = { version = "1.0", features = ["diff"] }
+
+[dependencies]
+proc-macro2 = "1.0"
+proc-macro-error = "1.0"
+syn = "2.0"
+quote = "1.0"

nac3core/nac3core_derive/src/lib.rs

@@ -0,0 +1,320 @@
+use proc_macro::TokenStream;
+use proc_macro_error::{abort, proc_macro_error};
+use quote::quote;
+use syn::{
+    parse_macro_input, spanned::Spanned, Data, DataStruct, Expr, ExprField, ExprMethodCall,
+    ExprPath, GenericArgument, Ident, LitStr, Path, PathArguments, Type, TypePath,
+};
+
+/// Extracts all generic arguments of a [`Type`] into a [`Vec`].
+///
+/// Returns [`Some`] of a possibly-empty [`Vec`] if the path of `ty` matches with
+/// `expected_ty_name`, otherwise returns [`None`].
+fn extract_generic_args(expected_ty_name: &'static str, ty: &Type) -> Option<Vec<GenericArgument>> {
+    let Type::Path(TypePath { qself: None, path, .. }) = ty else {
+        return None;
+    };
+
+    let segments = &path.segments;
+    if segments.len() != 1 {
+        return None;
+    };
+
+    let segment = segments.iter().next().unwrap();
+    if segment.ident != expected_ty_name {
+        return None;
+    }
+
+    let PathArguments::AngleBracketed(path_args) = &segment.arguments else {
+        return Some(Vec::new());
+    };
+    let args = &path_args.args;
+
+    Some(args.iter().cloned().collect::<Vec<_>>())
+}
+
+/// Maps a `path` matching one of the `target_idents` into the `replacement` [`Ident`].
+fn map_path_to_ident(path: &Path, target_idents: &[&str], replacement: &str) -> Option<Ident> {
+    path.require_ident()
+        .ok()
+        .filter(|ident| target_idents.iter().any(|target| ident == target))
+        .map(|ident| Ident::new(replacement, ident.span()))
+}
+
+/// Extracts the left-hand side of a dot-expression.
+fn extract_dot_operand(expr: &Expr) -> Option<&Expr> {
+    match expr {
+        Expr::MethodCall(ExprMethodCall { receiver: operand, .. })
+        | Expr::Field(ExprField { base: operand, .. }) => Some(operand),
+        _ => None,
+    }
+}
+
+/// Replaces the top-level receiver of a dot-expression with an [`Ident`], returning `Some(&mut expr)` if the
+/// replacement is performed.
+///
+/// The top-level receiver is the left-most receiver expression, e.g. the top-level receiver of `a.b.c.foo()` is `a`.
+fn replace_top_level_receiver(expr: &mut Expr, ident: Ident) -> Option<&mut Expr> {
+    if let Expr::MethodCall(ExprMethodCall { receiver: operand, .. })
+    | Expr::Field(ExprField { base: operand, .. }) = expr
+    {
+        return if extract_dot_operand(operand).is_some() {
+            if replace_top_level_receiver(operand, ident).is_some() {
+                Some(expr)
+            } else {
+                None
+            }
+        } else {
+            *operand = Box::new(Expr::Path(ExprPath {
+                attrs: Vec::default(),
+                qself: None,
+                path: ident.into(),
+            }));
+
+            Some(expr)
+        };
+    }
+
+    None
+}
+
+/// Iterates all operands to the left-hand side of the `.` of an [expression][`Expr`], i.e. the container operand of all
+/// [`Expr::Field`] and the receiver operand of all [`Expr::MethodCall`].
+///
+/// The iterator will return the operand expressions in reverse order of appearance. For example, `a.b.c.func()` will
+/// return `vec![c, b, a]`.
+fn iter_dot_operands(expr: &Expr) -> impl Iterator<Item = &Expr> {
+    let mut o = extract_dot_operand(expr);
+
+    std::iter::from_fn(move || {
+        let this = o;
+        o = o.as_ref().and_then(|o| extract_dot_operand(o));
+
+        this
+    })
+}
+
+/// Normalizes a value expression for use when creating an instance of this structure, returning a
+/// [`proc_macro2::TokenStream`] of tokens representing the normalized expression.
+fn normalize_value_expr(expr: &Expr) -> proc_macro2::TokenStream {
+    match &expr {
+        Expr::Path(ExprPath { qself: None, path, .. }) => {
+            if let Some(ident) = map_path_to_ident(path, &["usize", "size_t"], "llvm_usize") {
+                quote! { #ident }
+            } else {
+                abort!(
+                    path,
+                    format!(
+                        "Expected one of `size_t`, `usize`, or an implicit call expression in #[value_type(...)], found {}", 
+                        quote!(#expr).to_string(),
+                    )
+                )
+            }
+        }
+
+        Expr::Call(_) => {
+            quote! { ctx.#expr }
+        }
+
+        Expr::MethodCall(_) => {
+            let base_receiver = iter_dot_operands(expr).last();
+
+            match base_receiver {
+                // `usize.{...}`, `size_t.{...}` -> Rewrite the identifiers to `llvm_usize`
+                Some(Expr::Path(ExprPath { qself: None, path, .. }))
+                    if map_path_to_ident(path, &["usize", "size_t"], "llvm_usize").is_some() =>
+                {
+                    let ident =
+                        map_path_to_ident(path, &["usize", "size_t"], "llvm_usize").unwrap();
+
+                    let mut expr = expr.clone();
+                    let expr = replace_top_level_receiver(&mut expr, ident).unwrap();
+
+                    quote!(#expr)
+                }
+
+                // `ctx.{...}`, `context.{...}` -> Rewrite the identifiers to `ctx`
+                Some(Expr::Path(ExprPath { qself: None, path, .. }))
+                    if map_path_to_ident(path, &["ctx", "context"], "ctx").is_some() =>
+                {
+                    let ident = map_path_to_ident(path, &["ctx", "context"], "ctx").unwrap();
+
+                    let mut expr = expr.clone();
+                    let expr = replace_top_level_receiver(&mut expr, ident).unwrap();
+
+                    quote!(#expr)
+                }
+
+                // No reserved identifier prefix -> Prepend `ctx.` to the entire expression
+                _ => quote! { ctx.#expr },
+            }
+        }
+
+        _ => {
+            abort!(
+                expr,
+                format!(
+                    "Expected one of `size_t`, `usize`, or an implicit call expression in #[value_type(...)], found {}", 
+                    quote!(#expr).to_string(),
+                )
+            )
+        }
+    }
+}
+
+/// Derives an implementation of `codegen::types::structure::StructFields`.
+///
+/// The benefit of using `#[derive(StructFields)]` is that all index- or order-dependent logic required by
+/// `impl StructFields` is automatically generated by this implementation, including the field index as required by
+/// `StructField::new` and the fields as returned by `StructFields::to_vec`.
+///
+/// # Prerequisites
+///
+/// In order to derive from [`StructFields`], you must implement (or derive) [`Eq`] and [`Copy`] as required by
+/// `StructFields`.
+///
+/// Moreover, `#[derive(StructFields)]` can only be used for `struct`s with named fields, and may only contain fields
+/// with either `StructField` or [`PhantomData`] types.
+///
+/// # Attributes for [`StructFields`]
+///
+/// Each `StructField` field must be declared with the `#[value_type(...)]` attribute. The argument of `value_type`
+/// accepts one of the following:
+///
+/// - An expression returning an instance of `inkwell::types::BasicType` (with or without the receiver `ctx`/`context`).
+/// For example, `context.i8_type()`, `ctx.i8_type()`, and `i8_type()` all refer to `i8`.
+/// - The reserved identifiers `usize` and `size_t` referring to an `inkwell::types::IntType` of the platform-dependent
+/// integer size. `usize` and `size_t` can also be used as the receiver to other method calls, e.g.
+/// `usize.array_type(3)`.
+///
+/// # Example
+///
+/// The following is an example of an LLVM slice implemented using `#[derive(StructFields)]`.
+///
+/// ```rust,ignore
+/// use nac3core::{
+///     codegen::types::structure::StructField,
+///     inkwell::{
+///         values::{IntValue, PointerValue},
+///         AddressSpace,
+///     },
+/// };
+/// use nac3core_derive::StructFields;
+///
+/// // All classes that implement StructFields must also implement Eq and Copy
+/// #[derive(PartialEq, Eq, Clone, Copy, StructFields)]
+/// pub struct SliceValue<'ctx> {
+///     // Declares ptr have a value type of i8*
+///     //
+///     // Can also be written as `ctx.i8_type().ptr_type(...)` or `context.i8_type().ptr_type(...)`
+///     #[value_type(i8_type().ptr_type(AddressSpace::default()))]
+///     ptr: StructField<'ctx, PointerValue<'ctx>>,
+///
+///     // Declares len have a value type of usize, depending on the target compilation platform
+///     #[value_type(usize)]
+///     len: StructField<'ctx, IntValue<'ctx>>,
+/// }
+/// ```
+#[proc_macro_derive(StructFields, attributes(value_type))]
+#[proc_macro_error]
+pub fn derive(input: TokenStream) -> TokenStream {
+    let input = parse_macro_input!(input as syn::DeriveInput);
+    let ident = &input.ident;
+
+    let Data::Struct(DataStruct { fields, .. }) = &input.data else {
+        abort!(input, "Only structs with named fields are supported");
+    };
+    if let Err(err_span) =
+        fields
+            .iter()
+            .try_for_each(|field| if field.ident.is_some() { Ok(()) } else { Err(field.span()) })
+    {
+        abort!(err_span, "Only structs with named fields are supported");
+    };
+
+    // Check if struct<'ctx>
+    if input.generics.params.len() != 1 {
+        abort!(input.generics, "Expected exactly 1 generic parameter")
+    }
+
+    let phantom_info = fields
+        .iter()
+        .filter(|field| extract_generic_args("PhantomData", &field.ty).is_some())
+        .map(|field| field.ident.as_ref().unwrap())
+        .cloned()
+        .collect::<Vec<_>>();
+
+    let field_info = fields
+        .iter()
+        .filter(|field| extract_generic_args("PhantomData", &field.ty).is_none())
+        .map(|field| {
+            let ident = field.ident.as_ref().unwrap();
+            let ty = &field.ty;
+
+            let Some(_) = extract_generic_args("StructField", ty) else {
+                abort!(field, "Only StructField and PhantomData are allowed")
+            };
+
+            let attrs = &field.attrs;
+            let Some(value_type_attr) =
+                attrs.iter().find(|attr| attr.path().is_ident("value_type"))
+            else {
+                abort!(field, "Expected #[value_type(...)] attribute for field");
+            };
+
+            let Ok(value_type_expr) = value_type_attr.parse_args::<Expr>() else {
+                abort!(value_type_attr, "Expected expression in #[value_type(...)]");
+            };
+
+            let value_expr_toks = normalize_value_expr(&value_type_expr);
+
+            (ident.clone(), value_expr_toks)
+        })
+        .collect::<Vec<_>>();
+
+    // `<*>::new` impl of `StructField` and `PhantomData` for `StructFields::new`
+    let phantoms_create = phantom_info
+        .iter()
+        .map(|id| quote! { #id: ::std::marker::PhantomData })
+        .collect::<Vec<_>>();
+    let fields_create = field_info
+        .iter()
+        .map(|(id, ty)| {
+            let id_lit = LitStr::new(&id.to_string(), id.span());
+            quote! {
+                #id: ::nac3core::codegen::types::structure::StructField::create(
+                    &mut counter,
+                    #id_lit,
+                    #ty,
+                )
+            }
+        })
+        .collect::<Vec<_>>();
+
+    // `.into()` impl of `StructField` for `StructFields::to_vec`
+    let fields_into =
+        field_info.iter().map(|(id, _)| quote! { self.#id.into() }).collect::<Vec<_>>();
+
+    let impl_block = quote! {
+        impl<'ctx> ::nac3core::codegen::types::structure::StructFields<'ctx> for #ident<'ctx> {
+            fn new(ctx: impl ::nac3core::inkwell::context::AsContextRef<'ctx>, llvm_usize: ::nac3core::inkwell::types::IntType<'ctx>) -> Self {
+                let ctx = unsafe { ::nac3core::inkwell::context::ContextRef::new(ctx.as_ctx_ref()) };
+
+                let mut counter = ::nac3core::codegen::types::structure::FieldIndexCounter::default();
+
+                #ident {
+                    #(#fields_create),*
+                    #(#phantoms_create),*
+                }
+            }
+
+            fn to_vec(&self) -> ::std::vec::Vec<(&'static str, ::nac3core::inkwell::types::BasicTypeEnum<'ctx>)> {
+                vec![
+                    #(#fields_into),*
+                ]
+            }
+        }
+    };
+
+    impl_block.into()
+}

nac3core/nac3core_derive/tests/structfields_empty.rs

@@ -0,0 +1,9 @@
+use nac3core_derive::StructFields;
+use std::marker::PhantomData;
+
+#[derive(PartialEq, Eq, Clone, Copy, StructFields)]
+pub struct EmptyValue<'ctx> {
+    _phantom: PhantomData<&'ctx ()>,
+}
+
+fn main() {}

nac3core/nac3core_derive/tests/structfields_ndarray.rs

@@ -0,0 +1,20 @@
+use nac3core::{
+    codegen::types::structure::StructField,
+    inkwell::{
+        values::{IntValue, PointerValue},
+        AddressSpace,
+    },
+};
+use nac3core_derive::StructFields;
+
+#[derive(PartialEq, Eq, Clone, Copy, StructFields)]
+pub struct NDArrayValue<'ctx> {
+    #[value_type(usize)]
+    ndims: StructField<'ctx, IntValue<'ctx>>,
+    #[value_type(usize.ptr_type(AddressSpace::default()))]
+    shape: StructField<'ctx, PointerValue<'ctx>>,
+    #[value_type(i8_type().ptr_type(AddressSpace::default()))]
+    data: StructField<'ctx, PointerValue<'ctx>>,
+}
+
+fn main() {}

nac3core/nac3core_derive/tests/structfields_slice.rs

@@ -0,0 +1,18 @@
+use nac3core::{
+    codegen::types::structure::StructField,
+    inkwell::{
+        values::{IntValue, PointerValue},
+        AddressSpace,
+    },
+};
+use nac3core_derive::StructFields;
+
+#[derive(PartialEq, Eq, Clone, Copy, StructFields)]
+pub struct SliceValue<'ctx> {
+    #[value_type(i8_type().ptr_type(AddressSpace::default()))]
+    ptr: StructField<'ctx, PointerValue<'ctx>>,
+    #[value_type(usize)]
+    len: StructField<'ctx, IntValue<'ctx>>,
+}
+
+fn main() {}

nac3core/nac3core_derive/tests/structfields_slice_context.rs

@@ -0,0 +1,18 @@
+use nac3core::{
+    codegen::types::structure::StructField,
+    inkwell::{
+        values::{IntValue, PointerValue},
+        AddressSpace,
+    },
+};
+use nac3core_derive::StructFields;
+
+#[derive(PartialEq, Eq, Clone, Copy, StructFields)]
+pub struct SliceValue<'ctx> {
+    #[value_type(context.i8_type().ptr_type(AddressSpace::default()))]
+    ptr: StructField<'ctx, PointerValue<'ctx>>,
+    #[value_type(usize)]
+    len: StructField<'ctx, IntValue<'ctx>>,
+}
+
+fn main() {}

nac3core/nac3core_derive/tests/structfields_slice_ctx.rs

@@ -0,0 +1,18 @@
+use nac3core::{
+    codegen::types::structure::StructField,
+    inkwell::{
+        values::{IntValue, PointerValue},
+        AddressSpace,
+    },
+};
+use nac3core_derive::StructFields;
+
+#[derive(PartialEq, Eq, Clone, Copy, StructFields)]
+pub struct SliceValue<'ctx> {
+    #[value_type(ctx.i8_type().ptr_type(AddressSpace::default()))]
+    ptr: StructField<'ctx, PointerValue<'ctx>>,
+    #[value_type(usize)]
+    len: StructField<'ctx, IntValue<'ctx>>,
+}
+
+fn main() {}

nac3core/nac3core_derive/tests/structfields_slice_sizet.rs

@@ -0,0 +1,18 @@
+use nac3core::{
+    codegen::types::structure::StructField,
+    inkwell::{
+        values::{IntValue, PointerValue},
+        AddressSpace,
+    },
+};
+use nac3core_derive::StructFields;
+
+#[derive(PartialEq, Eq, Clone, Copy, StructFields)]
+pub struct SliceValue<'ctx> {
+    #[value_type(i8_type().ptr_type(AddressSpace::default()))]
+    ptr: StructField<'ctx, PointerValue<'ctx>>,
+    #[value_type(size_t)]
+    len: StructField<'ctx, IntValue<'ctx>>,
+}
+
+fn main() {}

nac3core/nac3core_derive/tests/structfields_test.rs

@@ -0,0 +1,10 @@
+#[test]
+fn test_parse_empty() {
+    let t = trybuild::TestCases::new();
+    t.pass("tests/structfields_empty.rs");
+    t.pass("tests/structfields_slice.rs");
+    t.pass("tests/structfields_slice_ctx.rs");
+    t.pass("tests/structfields_slice_context.rs");
+    t.pass("tests/structfields_slice_sizet.rs");
+    t.pass("tests/structfields_ndarray.rs");
+}

nac3core/src/codegen/concrete_type.rs

@@ -1,3 +1,9 @@
+use std::collections::HashMap;
+
+use indexmap::IndexMap;
+
+use nac3parser::ast::StrRef;
+
 use crate::{
     symbol_resolver::SymbolValue,
     toplevel::DefinitionId,
@@ -9,10 +15,6 @@ use crate::{
     },
 };
 
-use indexmap::IndexMap;
-use nac3parser::ast::StrRef;
-use std::collections::HashMap;
-
 pub struct ConcreteTypeStore {
     store: Vec<ConcreteTypeEnum>,
 }
@@ -25,6 +27,7 @@ pub struct ConcreteFuncArg {
     pub name: StrRef,
     pub ty: ConcreteType,
     pub default_value: Option<SymbolValue>,
+    pub is_vararg: bool,
 }
 
 #[derive(Clone, Debug)]
@@ -46,6 +49,7 @@ pub enum ConcreteTypeEnum {
     TPrimitive(Primitive),
     TTuple {
         ty: Vec<ConcreteType>,
+        is_vararg_ctx: bool,
     },
     TObj {
         obj_id: DefinitionId,
@@ -102,8 +106,16 @@ impl ConcreteTypeStore {
                 .iter()
                 .map(|arg| ConcreteFuncArg {
                     name: arg.name,
-                    ty: self.from_unifier_type(unifier, primitives, arg.ty, cache),
+                    ty: if arg.is_vararg {
+                        let tuple_ty = unifier
+                            .add_ty(TypeEnum::TTuple { ty: vec![arg.ty], is_vararg_ctx: true });
+
+                        self.from_unifier_type(unifier, primitives, tuple_ty, cache)
+                    } else {
+                        self.from_unifier_type(unifier, primitives, arg.ty, cache)
+                    },
                     default_value: arg.default_value.clone(),
+                    is_vararg: arg.is_vararg,
                 })
                 .collect(),
             ret: self.from_unifier_type(unifier, primitives, signature.ret, cache),
@@ -158,11 +170,12 @@ impl ConcreteTypeStore {
             cache.insert(ty, None);
             let ty_enum = unifier.get_ty(ty);
             let result = match &*ty_enum {
-                TypeEnum::TTuple { ty } => ConcreteTypeEnum::TTuple {
+                TypeEnum::TTuple { ty, is_vararg_ctx } => ConcreteTypeEnum::TTuple {
                     ty: ty
                         .iter()
                         .map(|t| self.from_unifier_type(unifier, primitives, *t, cache))
                         .collect(),
+                    is_vararg_ctx: *is_vararg_ctx,
                 },
                 TypeEnum::TObj { obj_id, fields, params } => ConcreteTypeEnum::TObj {
                     obj_id: *obj_id,
@@ -248,11 +261,12 @@ impl ConcreteTypeStore {
                 *cache.get_mut(&cty).unwrap() = Some(ty);
                 return ty;
             }
-            ConcreteTypeEnum::TTuple { ty } => TypeEnum::TTuple {
+            ConcreteTypeEnum::TTuple { ty, is_vararg_ctx } => TypeEnum::TTuple {
                 ty: ty
                     .iter()
                     .map(|cty| self.to_unifier_type(unifier, primitives, *cty, cache))
                     .collect(),
+                is_vararg_ctx: *is_vararg_ctx,
             },
             ConcreteTypeEnum::TVirtual { ty } => {
                 TypeEnum::TVirtual { ty: self.to_unifier_type(unifier, primitives, *ty, cache) }
@@ -277,6 +291,7 @@ impl ConcreteTypeStore {
                         name: arg.name,
                         ty: self.to_unifier_type(unifier, primitives, arg.ty, cache),
                         default_value: arg.default_value.clone(),
+                        is_vararg: false,
                     })
                     .collect(),
                 ret: self.to_unifier_type(unifier, primitives, *ret, cache),

nac3core/src/codegen/extern_fns.rs

@@ -1,8 +1,10 @@
-use inkwell::attributes::{Attribute, AttributeLoc};
-use inkwell::values::{BasicValueEnum, CallSiteValue, FloatValue, IntValue};
+use inkwell::{
+    attributes::{Attribute, AttributeLoc},
+    values::{BasicValueEnum, CallSiteValue, FloatValue, IntValue},
+};
 use itertools::Either;
 
-use crate::codegen::CodeGenContext;
+use super::CodeGenContext;
 
 /// Macro to generate extern function
 /// Both function return type and function parameter type are `FloatValue`
@@ -13,8 +15,8 @@ use crate::codegen::CodeGenContext;
 /// * `$extern_fn:literal`: Name of underlying extern function
 ///
 /// Optional Arguments:
-/// * `$(,$attributes:literal)*)`: Attributes linked with the extern function
-/// The default attributes are "mustprogress", "nofree", "nounwind", "willreturn", and "writeonly"
+/// * `$(,$attributes:literal)*)`: Attributes linked with the extern function.
+///   The default attributes are "mustprogress", "nofree", "nounwind", "willreturn", and "writeonly".
 ///   These will be used unless other attributes are specified
 /// * `$(,$args:ident)*`: Operands of the extern function
 ///   The data type of these operands will be set to `FloatValue`
@@ -130,3 +132,62 @@ pub fn call_ldexp<'ctx>(
         .map(Either::unwrap_left)
         .unwrap()
 }
+
+/// Macro to generate `np_linalg` and `sp_linalg` functions
+/// The function takes as input `NDArray` and returns ()
+///
+/// Arguments:
+/// * `$fn_name:ident`: The identifier of the rust function to be generated
+/// * `$extern_fn:literal`: Name of underlying extern function
+/// * (2/3/4): Number of `NDArray` that function takes as input
+///
+/// Note:
+/// The operands and resulting `NDArray` are both passed as input to the funcion
+/// It is the responsibility of caller to ensure that output `NDArray` is properly allocated on stack
+/// The function changes the content of the output `NDArray` in-place
+macro_rules! generate_linalg_extern_fn {
+    ($fn_name:ident, $extern_fn:literal, 2) => {
+        generate_linalg_extern_fn!($fn_name, $extern_fn, mat1, mat2);
+    };
+    ($fn_name:ident, $extern_fn:literal, 3) => {
+        generate_linalg_extern_fn!($fn_name, $extern_fn, mat1, mat2, mat3);
+    };
+    ($fn_name:ident, $extern_fn:literal, 4) => {
+        generate_linalg_extern_fn!($fn_name, $extern_fn, mat1, mat2, mat3, mat4);
+    };
+    ($fn_name:ident, $extern_fn:literal $(,$input_matrix:ident)*) => {
+        #[doc = concat!("Invokes the linalg `", stringify!($extern_fn), " function." )]
+        pub fn $fn_name<'ctx>(
+            ctx: &mut CodeGenContext<'ctx, '_>
+            $(,$input_matrix: BasicValueEnum<'ctx>)*,
+            name: Option<&str>,
+        ){
+            const FN_NAME: &str = $extern_fn;
+            let extern_fn = ctx.module.get_function(FN_NAME).unwrap_or_else(|| {
+                let fn_type = ctx.ctx.void_type().fn_type(&[$($input_matrix.get_type().into()),*], false);
+
+                let func = ctx.module.add_function(FN_NAME, fn_type, None);
+                for attr in ["mustprogress", "nofree", "nounwind", "willreturn", "writeonly"] {
+                    func.add_attribute(
+                        AttributeLoc::Function,
+                        ctx.ctx.create_enum_attribute(Attribute::get_named_enum_kind_id(attr), 0),
+                    );
+                }
+                func
+            });
+
+            ctx.builder.build_call(extern_fn, &[$($input_matrix.into(),)*], name.unwrap_or_default()).unwrap();
+        }
+    };
+}
+
+generate_linalg_extern_fn!(call_np_linalg_cholesky, "np_linalg_cholesky", 2);
+generate_linalg_extern_fn!(call_np_linalg_qr, "np_linalg_qr", 3);
+generate_linalg_extern_fn!(call_np_linalg_svd, "np_linalg_svd", 4);
+generate_linalg_extern_fn!(call_np_linalg_inv, "np_linalg_inv", 2);
+generate_linalg_extern_fn!(call_np_linalg_pinv, "np_linalg_pinv", 2);
+generate_linalg_extern_fn!(call_np_linalg_matrix_power, "np_linalg_matrix_power", 3);
+generate_linalg_extern_fn!(call_np_linalg_det, "np_linalg_det", 2);
+generate_linalg_extern_fn!(call_sp_linalg_lu, "sp_linalg_lu", 3);
+generate_linalg_extern_fn!(call_sp_linalg_schur, "sp_linalg_schur", 3);
+generate_linalg_extern_fn!(call_sp_linalg_hessenberg, "sp_linalg_hessenberg", 3);

nac3core/src/codegen/generator.rs

@@ -1,16 +1,18 @@
-use crate::{
-    codegen::{bool_to_i1, bool_to_i8, classes::ArraySliceValue, expr::*, stmt::*, CodeGenContext},
-    symbol_resolver::ValueEnum,
-    toplevel::{DefinitionId, TopLevelDef},
-    typecheck::typedef::{FunSignature, Type},
-};
 use inkwell::{
     context::Context,
     types::{BasicTypeEnum, IntType},
     values::{BasicValueEnum, IntValue, PointerValue},
 };
+
 use nac3parser::ast::{Expr, Stmt, StrRef};
 
+use super::{bool_to_i1, bool_to_i8, expr::*, stmt::*, values::ArraySliceValue, CodeGenContext};
+use crate::{
+    symbol_resolver::ValueEnum,
+    toplevel::{DefinitionId, TopLevelDef},
+    typecheck::typedef::{FunSignature, Type},
+};
+
 pub trait CodeGenerator {
     /// Return the module name for the code generator.
     fn get_name(&self) -> &str;
@@ -57,6 +59,7 @@ pub trait CodeGenerator {
     /// - fun: Function signature, definition ID and the substitution key.
     /// - params: Function parameters. Note that this does not include the object even if the
     ///   function is a class method.
+    ///
     /// Note that this function should check if the function is generated in another thread (due to
     /// possible race condition), see the default implementation for an example.
     fn gen_func_instance<'ctx>(
@@ -123,11 +126,45 @@ pub trait CodeGenerator {
         ctx: &mut CodeGenContext<'ctx, '_>,
         target: &Expr<Option<Type>>,
         value: ValueEnum<'ctx>,
+        value_ty: Type,
     ) -> Result<(), String>
     where
         Self: Sized,
     {
-        gen_assign(self, ctx, target, value)
+        gen_assign(self, ctx, target, value, value_ty)
+    }
+
+    /// Generate code for an assignment expression where LHS is a `"target_list"`.
+    ///
+    /// See <https://docs.python.org/3/reference/simple_stmts.html#assignment-statements>.
+    fn gen_assign_target_list<'ctx>(
+        &mut self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        targets: &Vec<Expr<Option<Type>>>,
+        value: ValueEnum<'ctx>,
+        value_ty: Type,
+    ) -> Result<(), String>
+    where
+        Self: Sized,
+    {
+        gen_assign_target_list(self, ctx, targets, value, value_ty)
+    }
+
+    /// Generate code for an item assignment.
+    ///
+    /// i.e., `target[key] = value`
+    fn gen_setitem<'ctx>(
+        &mut self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        target: &Expr<Option<Type>>,
+        key: &Expr<Option<Type>>,
+        value: ValueEnum<'ctx>,
+        value_ty: Type,
+    ) -> Result<(), String>
+    where
+        Self: Sized,
+    {
+        gen_setitem(self, ctx, target, key, value, value_ty)
     }
 
     /// Generate code for a while expression.

nac3core/src/codegen/irrt/list.rs

@@ -0,0 +1,162 @@
+use inkwell::{
+    types::BasicTypeEnum,
+    values::{BasicValueEnum, CallSiteValue, IntValue},
+    AddressSpace, IntPredicate,
+};
+use itertools::Either;
+
+use super::calculate_len_for_slice_range;
+use crate::codegen::{
+    macros::codegen_unreachable,
+    values::{ArrayLikeValue, ListValue},
+    CodeGenContext, CodeGenerator,
+};
+
+/// This function handles 'end' **inclusively**.
+/// Order of tuples `assign_idx` and `value_idx` is ('start', 'end', 'step').
+/// Negative index should be handled before entering this function
+pub fn list_slice_assignment<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    ty: BasicTypeEnum<'ctx>,
+    dest_arr: ListValue<'ctx>,
+    dest_idx: (IntValue<'ctx>, IntValue<'ctx>, IntValue<'ctx>),
+    src_arr: ListValue<'ctx>,
+    src_idx: (IntValue<'ctx>, IntValue<'ctx>, IntValue<'ctx>),
+) {
+    let size_ty = generator.get_size_type(ctx.ctx);
+    let int8_ptr = ctx.ctx.i8_type().ptr_type(AddressSpace::default());
+    let int32 = ctx.ctx.i32_type();
+    let (fun_symbol, elem_ptr_type) = ("__nac3_list_slice_assign_var_size", int8_ptr);
+    let slice_assign_fun = {
+        let ty_vec = vec![
+            int32.into(),         // dest start idx
+            int32.into(),         // dest end idx
+            int32.into(),         // dest step
+            elem_ptr_type.into(), // dest arr ptr
+            int32.into(),         // dest arr len
+            int32.into(),         // src start idx
+            int32.into(),         // src end idx
+            int32.into(),         // src step
+            elem_ptr_type.into(), // src arr ptr
+            int32.into(),         // src arr len
+            int32.into(),         // size
+        ];
+        ctx.module.get_function(fun_symbol).unwrap_or_else(|| {
+            let fn_t = int32.fn_type(ty_vec.as_slice(), false);
+            ctx.module.add_function(fun_symbol, fn_t, None)
+        })
+    };
+
+    let zero = int32.const_zero();
+    let one = int32.const_int(1, false);
+    let dest_arr_ptr = dest_arr.data().base_ptr(ctx, generator);
+    let dest_arr_ptr =
+        ctx.builder.build_pointer_cast(dest_arr_ptr, elem_ptr_type, "dest_arr_ptr_cast").unwrap();
+    let dest_len = dest_arr.load_size(ctx, Some("dest.len"));
+    let dest_len = ctx.builder.build_int_truncate_or_bit_cast(dest_len, int32, "srclen32").unwrap();
+    let src_arr_ptr = src_arr.data().base_ptr(ctx, generator);
+    let src_arr_ptr =
+        ctx.builder.build_pointer_cast(src_arr_ptr, elem_ptr_type, "src_arr_ptr_cast").unwrap();
+    let src_len = src_arr.load_size(ctx, Some("src.len"));
+    let src_len = ctx.builder.build_int_truncate_or_bit_cast(src_len, int32, "srclen32").unwrap();
+
+    // index in bound and positive should be done
+    // assert if dest.step == 1 then len(src) <= len(dest) else len(src) == len(dest), and
+    // throw exception if not satisfied
+    let src_end = ctx
+        .builder
+        .build_select(
+            ctx.builder.build_int_compare(IntPredicate::SLT, src_idx.2, zero, "is_neg").unwrap(),
+            ctx.builder.build_int_sub(src_idx.1, one, "e_min_one").unwrap(),
+            ctx.builder.build_int_add(src_idx.1, one, "e_add_one").unwrap(),
+            "final_e",
+        )
+        .map(BasicValueEnum::into_int_value)
+        .unwrap();
+    let dest_end = ctx
+        .builder
+        .build_select(
+            ctx.builder.build_int_compare(IntPredicate::SLT, dest_idx.2, zero, "is_neg").unwrap(),
+            ctx.builder.build_int_sub(dest_idx.1, one, "e_min_one").unwrap(),
+            ctx.builder.build_int_add(dest_idx.1, one, "e_add_one").unwrap(),
+            "final_e",
+        )
+        .map(BasicValueEnum::into_int_value)
+        .unwrap();
+    let src_slice_len =
+        calculate_len_for_slice_range(generator, ctx, src_idx.0, src_end, src_idx.2);
+    let dest_slice_len =
+        calculate_len_for_slice_range(generator, ctx, dest_idx.0, dest_end, dest_idx.2);
+    let src_eq_dest = ctx
+        .builder
+        .build_int_compare(IntPredicate::EQ, src_slice_len, dest_slice_len, "slice_src_eq_dest")
+        .unwrap();
+    let src_slt_dest = ctx
+        .builder
+        .build_int_compare(IntPredicate::SLT, src_slice_len, dest_slice_len, "slice_src_slt_dest")
+        .unwrap();
+    let dest_step_eq_one = ctx
+        .builder
+        .build_int_compare(
+            IntPredicate::EQ,
+            dest_idx.2,
+            dest_idx.2.get_type().const_int(1, false),
+            "slice_dest_step_eq_one",
+        )
+        .unwrap();
+    let cond_1 = ctx.builder.build_and(dest_step_eq_one, src_slt_dest, "slice_cond_1").unwrap();
+    let cond = ctx.builder.build_or(src_eq_dest, cond_1, "slice_cond").unwrap();
+    ctx.make_assert(
+        generator,
+        cond,
+        "0:ValueError",
+        "attempt to assign sequence of size {0} to slice of size {1} with step size {2}",
+        [Some(src_slice_len), Some(dest_slice_len), Some(dest_idx.2)],
+        ctx.current_loc,
+    );
+
+    let new_len = {
+        let args = vec![
+            dest_idx.0.into(),   // dest start idx
+            dest_idx.1.into(),   // dest end idx
+            dest_idx.2.into(),   // dest step
+            dest_arr_ptr.into(), // dest arr ptr
+            dest_len.into(),     // dest arr len
+            src_idx.0.into(),    // src start idx
+            src_idx.1.into(),    // src end idx
+            src_idx.2.into(),    // src step
+            src_arr_ptr.into(),  // src arr ptr
+            src_len.into(),      // src arr len
+            {
+                let s = match ty {
+                    BasicTypeEnum::FloatType(t) => t.size_of(),
+                    BasicTypeEnum::IntType(t) => t.size_of(),
+                    BasicTypeEnum::PointerType(t) => t.size_of(),
+                    BasicTypeEnum::StructType(t) => t.size_of().unwrap(),
+                    _ => codegen_unreachable!(ctx),
+                };
+                ctx.builder.build_int_truncate_or_bit_cast(s, int32, "size").unwrap()
+            }
+            .into(),
+        ];
+        ctx.builder
+            .build_call(slice_assign_fun, args.as_slice(), "slice_assign")
+            .map(CallSiteValue::try_as_basic_value)
+            .map(|v| v.map_left(BasicValueEnum::into_int_value))
+            .map(Either::unwrap_left)
+            .unwrap()
+    };
+    // update length
+    let need_update =
+        ctx.builder.build_int_compare(IntPredicate::NE, new_len, dest_len, "need_update").unwrap();
+    let current = ctx.builder.get_insert_block().unwrap().get_parent().unwrap();
+    let update_bb = ctx.ctx.append_basic_block(current, "update");
+    let cont_bb = ctx.ctx.append_basic_block(current, "cont");
+    ctx.builder.build_conditional_branch(need_update, update_bb, cont_bb).unwrap();
+    ctx.builder.position_at_end(update_bb);
+    let new_len = ctx.builder.build_int_z_extend_or_bit_cast(new_len, size_ty, "new_len").unwrap();
+    dest_arr.store_size(ctx, generator, new_len);
+    ctx.builder.build_unconditional_branch(cont_bb).unwrap();
+    ctx.builder.position_at_end(cont_bb);
+}

nac3core/src/codegen/irrt/math.rs

@@ -0,0 +1,152 @@
+use inkwell::{
+    values::{BasicValueEnum, CallSiteValue, FloatValue, IntValue},
+    IntPredicate,
+};
+use itertools::Either;
+
+use crate::codegen::{
+    macros::codegen_unreachable,
+    {CodeGenContext, CodeGenerator},
+};
+
+// repeated squaring method adapted from GNU Scientific Library:
+// https://git.savannah.gnu.org/cgit/gsl.git/tree/sys/pow_int.c
+pub fn integer_power<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    base: IntValue<'ctx>,
+    exp: IntValue<'ctx>,
+    signed: bool,
+) -> IntValue<'ctx> {
+    let symbol = match (base.get_type().get_bit_width(), exp.get_type().get_bit_width(), signed) {
+        (32, 32, true) => "__nac3_int_exp_int32_t",
+        (64, 64, true) => "__nac3_int_exp_int64_t",
+        (32, 32, false) => "__nac3_int_exp_uint32_t",
+        (64, 64, false) => "__nac3_int_exp_uint64_t",
+        _ => codegen_unreachable!(ctx),
+    };
+    let base_type = base.get_type();
+    let pow_fun = ctx.module.get_function(symbol).unwrap_or_else(|| {
+        let fn_type = base_type.fn_type(&[base_type.into(), base_type.into()], false);
+        ctx.module.add_function(symbol, fn_type, None)
+    });
+    // throw exception when exp < 0
+    let ge_zero = ctx
+        .builder
+        .build_int_compare(
+            IntPredicate::SGE,
+            exp,
+            exp.get_type().const_zero(),
+            "assert_int_pow_ge_0",
+        )
+        .unwrap();
+    ctx.make_assert(
+        generator,
+        ge_zero,
+        "0:ValueError",
+        "integer power must be positive or zero",
+        [None, None, None],
+        ctx.current_loc,
+    );
+    ctx.builder
+        .build_call(pow_fun, &[base.into(), exp.into()], "call_int_pow")
+        .map(CallSiteValue::try_as_basic_value)
+        .map(|v| v.map_left(BasicValueEnum::into_int_value))
+        .map(Either::unwrap_left)
+        .unwrap()
+}
+
+/// Generates a call to `isinf` in IR. Returns an `i1` representing the result.
+pub fn call_isinf<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &CodeGenContext<'ctx, '_>,
+    v: FloatValue<'ctx>,
+) -> IntValue<'ctx> {
+    let intrinsic_fn = ctx.module.get_function("__nac3_isinf").unwrap_or_else(|| {
+        let fn_type = ctx.ctx.i32_type().fn_type(&[ctx.ctx.f64_type().into()], false);
+        ctx.module.add_function("__nac3_isinf", fn_type, None)
+    });
+
+    let ret = ctx
+        .builder
+        .build_call(intrinsic_fn, &[v.into()], "isinf")
+        .map(CallSiteValue::try_as_basic_value)
+        .map(|v| v.map_left(BasicValueEnum::into_int_value))
+        .map(Either::unwrap_left)
+        .unwrap();
+
+    generator.bool_to_i1(ctx, ret)
+}
+
+/// Generates a call to `isnan` in IR. Returns an `i1` representing the result.
+pub fn call_isnan<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &CodeGenContext<'ctx, '_>,
+    v: FloatValue<'ctx>,
+) -> IntValue<'ctx> {
+    let intrinsic_fn = ctx.module.get_function("__nac3_isnan").unwrap_or_else(|| {
+        let fn_type = ctx.ctx.i32_type().fn_type(&[ctx.ctx.f64_type().into()], false);
+        ctx.module.add_function("__nac3_isnan", fn_type, None)
+    });
+
+    let ret = ctx
+        .builder
+        .build_call(intrinsic_fn, &[v.into()], "isnan")
+        .map(CallSiteValue::try_as_basic_value)
+        .map(|v| v.map_left(BasicValueEnum::into_int_value))
+        .map(Either::unwrap_left)
+        .unwrap();
+
+    generator.bool_to_i1(ctx, ret)
+}
+
+/// Generates a call to `gamma` in IR. Returns an `f64` representing the result.
+pub fn call_gamma<'ctx>(ctx: &CodeGenContext<'ctx, '_>, v: FloatValue<'ctx>) -> FloatValue<'ctx> {
+    let llvm_f64 = ctx.ctx.f64_type();
+
+    let intrinsic_fn = ctx.module.get_function("__nac3_gamma").unwrap_or_else(|| {
+        let fn_type = llvm_f64.fn_type(&[llvm_f64.into()], false);
+        ctx.module.add_function("__nac3_gamma", fn_type, None)
+    });
+
+    ctx.builder
+        .build_call(intrinsic_fn, &[v.into()], "gamma")
+        .map(CallSiteValue::try_as_basic_value)
+        .map(|v| v.map_left(BasicValueEnum::into_float_value))
+        .map(Either::unwrap_left)
+        .unwrap()
+}
+
+/// Generates a call to `gammaln` in IR. Returns an `f64` representing the result.
+pub fn call_gammaln<'ctx>(ctx: &CodeGenContext<'ctx, '_>, v: FloatValue<'ctx>) -> FloatValue<'ctx> {
+    let llvm_f64 = ctx.ctx.f64_type();
+
+    let intrinsic_fn = ctx.module.get_function("__nac3_gammaln").unwrap_or_else(|| {
+        let fn_type = llvm_f64.fn_type(&[llvm_f64.into()], false);
+        ctx.module.add_function("__nac3_gammaln", fn_type, None)
+    });
+
+    ctx.builder
+        .build_call(intrinsic_fn, &[v.into()], "gammaln")
+        .map(CallSiteValue::try_as_basic_value)
+        .map(|v| v.map_left(BasicValueEnum::into_float_value))
+        .map(Either::unwrap_left)
+        .unwrap()
+}
+
+/// Generates a call to `j0` in IR. Returns an `f64` representing the result.
+pub fn call_j0<'ctx>(ctx: &CodeGenContext<'ctx, '_>, v: FloatValue<'ctx>) -> FloatValue<'ctx> {
+    let llvm_f64 = ctx.ctx.f64_type();
+
+    let intrinsic_fn = ctx.module.get_function("__nac3_j0").unwrap_or_else(|| {
+        let fn_type = llvm_f64.fn_type(&[llvm_f64.into()], false);
+        ctx.module.add_function("__nac3_j0", fn_type, None)
+    });
+
+    ctx.builder
+        .build_call(intrinsic_fn, &[v.into()], "j0")
+        .map(CallSiteValue::try_as_basic_value)
+        .map(|v| v.map_left(BasicValueEnum::into_float_value))
+        .map(Either::unwrap_left)
+        .unwrap()
+}

nac3core/src/codegen/irrt/mod.rs

@@ -1,30 +1,28 @@
-use crate::{typecheck::typedef::Type, util::SizeVariant};
-
-mod test;
-
-use super::{
-    classes::{
-        ArrayLikeIndexer, ArrayLikeValue, ArraySliceValue, ListValue, NDArrayValue, NpArrayType,
-        NpArrayValue, TypedArrayLikeAdapter, UntypedArrayLikeAccessor,
-    },
-    llvm_intrinsics, CodeGenContext, CodeGenerator,
-};
-use crate::codegen::classes::TypedArrayLikeAccessor;
-use crate::codegen::stmt::gen_for_callback_incrementing;
 use inkwell::{
     attributes::{Attribute, AttributeLoc},
     context::Context,
     memory_buffer::MemoryBuffer,
     module::Module,
-    types::{BasicType, BasicTypeEnum, FunctionType, IntType, PointerType},
-    values::{BasicValueEnum, CallSiteValue, FloatValue, FunctionValue, IntValue},
-    AddressSpace, IntPredicate,
+    values::{BasicValue, BasicValueEnum, IntValue},
+    IntPredicate,
 };
-use itertools::Either;
+
 use nac3parser::ast::Expr;
 
+use super::{CodeGenContext, CodeGenerator};
+use crate::{symbol_resolver::SymbolResolver, typecheck::typedef::Type};
+pub use list::*;
+pub use math::*;
+pub use ndarray::*;
+pub use slice::*;
+
+mod list;
+mod math;
+mod ndarray;
+mod slice;
+
 #[must_use]
-pub fn load_irrt(ctx: &Context) -> Module {
+pub fn load_irrt<'ctx>(ctx: &'ctx Context, symbol_resolver: &dyn SymbolResolver) -> Module<'ctx> {
     let bitcode_buf = MemoryBuffer::create_from_memory_range(
         include_bytes!(concat!(env!("OUT_DIR"), "/irrt.bc")),
         "irrt_bitcode_buffer",
@@ -40,91 +38,28 @@ pub fn load_irrt(ctx: &Context) -> Module {
         let function = irrt_mod.get_function(symbol).unwrap();
         function.add_attribute(AttributeLoc::Function, ctx.create_enum_attribute(inline_attr, 0));
     }
+
+    // Initialize all global `EXN_*` exception IDs in IRRT with the [`SymbolResolver`].
+    let exn_id_type = ctx.i32_type();
+    let errors = &[
+        ("EXN_INDEX_ERROR", "0:IndexError"),
+        ("EXN_VALUE_ERROR", "0:ValueError"),
+        ("EXN_ASSERTION_ERROR", "0:AssertionError"),
+        ("EXN_TYPE_ERROR", "0:TypeError"),
+    ];
+    for (irrt_name, symbol_name) in errors {
+        let exn_id = symbol_resolver.get_string_id(symbol_name);
+        let exn_id = exn_id_type.const_int(exn_id as u64, false).as_basic_value_enum();
+
+        let global = irrt_mod.get_global(irrt_name).unwrap_or_else(|| {
+            panic!("Exception symbol name '{irrt_name}' should exist in the IRRT LLVM module")
+        });
+        global.set_initializer(&exn_id);
+    }
+
     irrt_mod
 }
 
-// repeated squaring method adapted from GNU Scientific Library:
-// https://git.savannah.gnu.org/cgit/gsl.git/tree/sys/pow_int.c
-pub fn integer_power<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &mut G,
-    ctx: &mut CodeGenContext<'ctx, '_>,
-    base: IntValue<'ctx>,
-    exp: IntValue<'ctx>,
-    signed: bool,
-) -> IntValue<'ctx> {
-    let symbol = match (base.get_type().get_bit_width(), exp.get_type().get_bit_width(), signed) {
-        (32, 32, true) => "__nac3_int_exp_int32_t",
-        (64, 64, true) => "__nac3_int_exp_int64_t",
-        (32, 32, false) => "__nac3_int_exp_uint32_t",
-        (64, 64, false) => "__nac3_int_exp_uint64_t",
-        _ => unreachable!(),
-    };
-    let base_type = base.get_type();
-    let pow_fun = ctx.module.get_function(symbol).unwrap_or_else(|| {
-        let fn_type = base_type.fn_type(&[base_type.into(), base_type.into()], false);
-        ctx.module.add_function(symbol, fn_type, None)
-    });
-    // throw exception when exp < 0
-    let ge_zero = ctx
-        .builder
-        .build_int_compare(
-            IntPredicate::SGE,
-            exp,
-            exp.get_type().const_zero(),
-            "assert_int_pow_ge_0",
-        )
-        .unwrap();
-    ctx.make_assert(
-        generator,
-        ge_zero,
-        "0:ValueError",
-        "integer power must be positive or zero",
-        [None, None, None],
-        ctx.current_loc,
-    );
-    ctx.builder
-        .build_call(pow_fun, &[base.into(), exp.into()], "call_int_pow")
-        .map(CallSiteValue::try_as_basic_value)
-        .map(|v| v.map_left(BasicValueEnum::into_int_value))
-        .map(Either::unwrap_left)
-        .unwrap()
-}
-
-pub fn calculate_len_for_slice_range<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &mut G,
-    ctx: &mut CodeGenContext<'ctx, '_>,
-    start: IntValue<'ctx>,
-    end: IntValue<'ctx>,
-    step: IntValue<'ctx>,
-) -> IntValue<'ctx> {
-    const SYMBOL: &str = "__nac3_range_slice_len";
-    let len_func = ctx.module.get_function(SYMBOL).unwrap_or_else(|| {
-        let i32_t = ctx.ctx.i32_type();
-        let fn_t = i32_t.fn_type(&[i32_t.into(), i32_t.into(), i32_t.into()], false);
-        ctx.module.add_function(SYMBOL, fn_t, None)
-    });
-
-    // assert step != 0, throw exception if not
-    let not_zero = ctx
-        .builder
-        .build_int_compare(IntPredicate::NE, step, step.get_type().const_zero(), "range_step_ne")
-        .unwrap();
-    ctx.make_assert(
-        generator,
-        not_zero,
-        "0:ValueError",
-        "step must not be zero",
-        [None, None, None],
-        ctx.current_loc,
-    );
-    ctx.builder
-        .build_call(len_func, &[start.into(), end.into(), step.into()], "calc_len")
-        .map(CallSiteValue::try_as_basic_value)
-        .map(|v| v.map_left(BasicValueEnum::into_int_value))
-        .map(Either::unwrap_left)
-        .unwrap()
-}
-
 /// NOTE: the output value of the end index of this function should be compared ***inclusively***,
 /// because python allows `a[2::-1]`, whose semantic is `[a[2], a[1], a[0]]`, which is equivalent to
 /// NO numeric slice in python.
@@ -290,702 +225,3 @@ pub fn handle_slice_indices<'ctx, G: CodeGenerator>(
         }
     }))
 }
-
-/// this function allows index out of range, since python
-/// allows index out of range in slice (`a = [1,2,3]; a[1:10] == [2,3]`).
-pub fn handle_slice_index_bound<'ctx, G: CodeGenerator>(
-    i: &Expr<Option<Type>>,
-    ctx: &mut CodeGenContext<'ctx, '_>,
-    generator: &mut G,
-    length: IntValue<'ctx>,
-) -> Result<Option<IntValue<'ctx>>, String> {
-    const SYMBOL: &str = "__nac3_slice_index_bound";
-    let func = ctx.module.get_function(SYMBOL).unwrap_or_else(|| {
-        let i32_t = ctx.ctx.i32_type();
-        let fn_t = i32_t.fn_type(&[i32_t.into(), i32_t.into()], false);
-        ctx.module.add_function(SYMBOL, fn_t, None)
-    });
-
-    let i = if let Some(v) = generator.gen_expr(ctx, i)? {
-        v.to_basic_value_enum(ctx, generator, i.custom.unwrap())?
-    } else {
-        return Ok(None);
-    };
-    Ok(Some(
-        ctx.builder
-            .build_call(func, &[i.into(), length.into()], "bounded_ind")
-            .map(CallSiteValue::try_as_basic_value)
-            .map(|v| v.map_left(BasicValueEnum::into_int_value))
-            .map(Either::unwrap_left)
-            .unwrap(),
-    ))
-}
-
-/// This function handles 'end' **inclusively**.
-/// Order of tuples `assign_idx` and `value_idx` is ('start', 'end', 'step').
-/// Negative index should be handled before entering this function
-pub fn list_slice_assignment<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &mut G,
-    ctx: &mut CodeGenContext<'ctx, '_>,
-    ty: BasicTypeEnum<'ctx>,
-    dest_arr: ListValue<'ctx>,
-    dest_idx: (IntValue<'ctx>, IntValue<'ctx>, IntValue<'ctx>),
-    src_arr: ListValue<'ctx>,
-    src_idx: (IntValue<'ctx>, IntValue<'ctx>, IntValue<'ctx>),
-) {
-    let size_ty = generator.get_size_type(ctx.ctx);
-    let int8_ptr = ctx.ctx.i8_type().ptr_type(AddressSpace::default());
-    let int32 = ctx.ctx.i32_type();
-    let (fun_symbol, elem_ptr_type) = ("__nac3_list_slice_assign_var_size", int8_ptr);
-    let slice_assign_fun = {
-        let ty_vec = vec![
-            int32.into(),         // dest start idx
-            int32.into(),         // dest end idx
-            int32.into(),         // dest step
-            elem_ptr_type.into(), // dest arr ptr
-            int32.into(),         // dest arr len
-            int32.into(),         // src start idx
-            int32.into(),         // src end idx
-            int32.into(),         // src step
-            elem_ptr_type.into(), // src arr ptr
-            int32.into(),         // src arr len
-            int32.into(),         // size
-        ];
-        ctx.module.get_function(fun_symbol).unwrap_or_else(|| {
-            let fn_t = int32.fn_type(ty_vec.as_slice(), false);
-            ctx.module.add_function(fun_symbol, fn_t, None)
-        })
-    };
-
-    let zero = int32.const_zero();
-    let one = int32.const_int(1, false);
-    let dest_arr_ptr = dest_arr.data().base_ptr(ctx, generator);
-    let dest_arr_ptr =
-        ctx.builder.build_pointer_cast(dest_arr_ptr, elem_ptr_type, "dest_arr_ptr_cast").unwrap();
-    let dest_len = dest_arr.load_size(ctx, Some("dest.len"));
-    let dest_len = ctx.builder.build_int_truncate_or_bit_cast(dest_len, int32, "srclen32").unwrap();
-    let src_arr_ptr = src_arr.data().base_ptr(ctx, generator);
-    let src_arr_ptr =
-        ctx.builder.build_pointer_cast(src_arr_ptr, elem_ptr_type, "src_arr_ptr_cast").unwrap();
-    let src_len = src_arr.load_size(ctx, Some("src.len"));
-    let src_len = ctx.builder.build_int_truncate_or_bit_cast(src_len, int32, "srclen32").unwrap();
-
-    // index in bound and positive should be done
-    // assert if dest.step == 1 then len(src) <= len(dest) else len(src) == len(dest), and
-    // throw exception if not satisfied
-    let src_end = ctx
-        .builder
-        .build_select(
-            ctx.builder.build_int_compare(IntPredicate::SLT, src_idx.2, zero, "is_neg").unwrap(),
-            ctx.builder.build_int_sub(src_idx.1, one, "e_min_one").unwrap(),
-            ctx.builder.build_int_add(src_idx.1, one, "e_add_one").unwrap(),
-            "final_e",
-        )
-        .map(BasicValueEnum::into_int_value)
-        .unwrap();
-    let dest_end = ctx
-        .builder
-        .build_select(
-            ctx.builder.build_int_compare(IntPredicate::SLT, dest_idx.2, zero, "is_neg").unwrap(),
-            ctx.builder.build_int_sub(dest_idx.1, one, "e_min_one").unwrap(),
-            ctx.builder.build_int_add(dest_idx.1, one, "e_add_one").unwrap(),
-            "final_e",
-        )
-        .map(BasicValueEnum::into_int_value)
-        .unwrap();
-    let src_slice_len =
-        calculate_len_for_slice_range(generator, ctx, src_idx.0, src_end, src_idx.2);
-    let dest_slice_len =
-        calculate_len_for_slice_range(generator, ctx, dest_idx.0, dest_end, dest_idx.2);
-    let src_eq_dest = ctx
-        .builder
-        .build_int_compare(IntPredicate::EQ, src_slice_len, dest_slice_len, "slice_src_eq_dest")
-        .unwrap();
-    let src_slt_dest = ctx
-        .builder
-        .build_int_compare(IntPredicate::SLT, src_slice_len, dest_slice_len, "slice_src_slt_dest")
-        .unwrap();
-    let dest_step_eq_one = ctx
-        .builder
-        .build_int_compare(
-            IntPredicate::EQ,
-            dest_idx.2,
-            dest_idx.2.get_type().const_int(1, false),
-            "slice_dest_step_eq_one",
-        )
-        .unwrap();
-    let cond_1 = ctx.builder.build_and(dest_step_eq_one, src_slt_dest, "slice_cond_1").unwrap();
-    let cond = ctx.builder.build_or(src_eq_dest, cond_1, "slice_cond").unwrap();
-    ctx.make_assert(
-        generator,
-        cond,
-        "0:ValueError",
-        "attempt to assign sequence of size {0} to slice of size {1} with step size {2}",
-        [Some(src_slice_len), Some(dest_slice_len), Some(dest_idx.2)],
-        ctx.current_loc,
-    );
-
-    let new_len = {
-        let args = vec![
-            dest_idx.0.into(),   // dest start idx
-            dest_idx.1.into(),   // dest end idx
-            dest_idx.2.into(),   // dest step
-            dest_arr_ptr.into(), // dest arr ptr
-            dest_len.into(),     // dest arr len
-            src_idx.0.into(),    // src start idx
-            src_idx.1.into(),    // src end idx
-            src_idx.2.into(),    // src step
-            src_arr_ptr.into(),  // src arr ptr
-            src_len.into(),      // src arr len
-            {
-                let s = match ty {
-                    BasicTypeEnum::FloatType(t) => t.size_of(),
-                    BasicTypeEnum::IntType(t) => t.size_of(),
-                    BasicTypeEnum::PointerType(t) => t.size_of(),
-                    BasicTypeEnum::StructType(t) => t.size_of().unwrap(),
-                    _ => unreachable!(),
-                };
-                ctx.builder.build_int_truncate_or_bit_cast(s, int32, "size").unwrap()
-            }
-            .into(),
-        ];
-        ctx.builder
-            .build_call(slice_assign_fun, args.as_slice(), "slice_assign")
-            .map(CallSiteValue::try_as_basic_value)
-            .map(|v| v.map_left(BasicValueEnum::into_int_value))
-            .map(Either::unwrap_left)
-            .unwrap()
-    };
-    // update length
-    let need_update =
-        ctx.builder.build_int_compare(IntPredicate::NE, new_len, dest_len, "need_update").unwrap();
-    let current = ctx.builder.get_insert_block().unwrap().get_parent().unwrap();
-    let update_bb = ctx.ctx.append_basic_block(current, "update");
-    let cont_bb = ctx.ctx.append_basic_block(current, "cont");
-    ctx.builder.build_conditional_branch(need_update, update_bb, cont_bb).unwrap();
-    ctx.builder.position_at_end(update_bb);
-    let new_len = ctx.builder.build_int_z_extend_or_bit_cast(new_len, size_ty, "new_len").unwrap();
-    dest_arr.store_size(ctx, generator, new_len);
-    ctx.builder.build_unconditional_branch(cont_bb).unwrap();
-    ctx.builder.position_at_end(cont_bb);
-}
-
-/// Generates a call to `isinf` in IR. Returns an `i1` representing the result.
-pub fn call_isinf<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &mut G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    v: FloatValue<'ctx>,
-) -> IntValue<'ctx> {
-    let intrinsic_fn = ctx.module.get_function("__nac3_isinf").unwrap_or_else(|| {
-        let fn_type = ctx.ctx.i32_type().fn_type(&[ctx.ctx.f64_type().into()], false);
-        ctx.module.add_function("__nac3_isinf", fn_type, None)
-    });
-
-    let ret = ctx
-        .builder
-        .build_call(intrinsic_fn, &[v.into()], "isinf")
-        .map(CallSiteValue::try_as_basic_value)
-        .map(|v| v.map_left(BasicValueEnum::into_int_value))
-        .map(Either::unwrap_left)
-        .unwrap();
-
-    generator.bool_to_i1(ctx, ret)
-}
-
-/// Generates a call to `isnan` in IR. Returns an `i1` representing the result.
-pub fn call_isnan<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &mut G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    v: FloatValue<'ctx>,
-) -> IntValue<'ctx> {
-    let intrinsic_fn = ctx.module.get_function("__nac3_isnan").unwrap_or_else(|| {
-        let fn_type = ctx.ctx.i32_type().fn_type(&[ctx.ctx.f64_type().into()], false);
-        ctx.module.add_function("__nac3_isnan", fn_type, None)
-    });
-
-    let ret = ctx
-        .builder
-        .build_call(intrinsic_fn, &[v.into()], "isnan")
-        .map(CallSiteValue::try_as_basic_value)
-        .map(|v| v.map_left(BasicValueEnum::into_int_value))
-        .map(Either::unwrap_left)
-        .unwrap();
-
-    generator.bool_to_i1(ctx, ret)
-}
-
-/// Generates a call to `gamma` in IR. Returns an `f64` representing the result.
-pub fn call_gamma<'ctx>(ctx: &CodeGenContext<'ctx, '_>, v: FloatValue<'ctx>) -> FloatValue<'ctx> {
-    let llvm_f64 = ctx.ctx.f64_type();
-
-    let intrinsic_fn = ctx.module.get_function("__nac3_gamma").unwrap_or_else(|| {
-        let fn_type = llvm_f64.fn_type(&[llvm_f64.into()], false);
-        ctx.module.add_function("__nac3_gamma", fn_type, None)
-    });
-
-    ctx.builder
-        .build_call(intrinsic_fn, &[v.into()], "gamma")
-        .map(CallSiteValue::try_as_basic_value)
-        .map(|v| v.map_left(BasicValueEnum::into_float_value))
-        .map(Either::unwrap_left)
-        .unwrap()
-}
-
-/// Generates a call to `gammaln` in IR. Returns an `f64` representing the result.
-pub fn call_gammaln<'ctx>(ctx: &CodeGenContext<'ctx, '_>, v: FloatValue<'ctx>) -> FloatValue<'ctx> {
-    let llvm_f64 = ctx.ctx.f64_type();
-
-    let intrinsic_fn = ctx.module.get_function("__nac3_gammaln").unwrap_or_else(|| {
-        let fn_type = llvm_f64.fn_type(&[llvm_f64.into()], false);
-        ctx.module.add_function("__nac3_gammaln", fn_type, None)
-    });
-
-    ctx.builder
-        .build_call(intrinsic_fn, &[v.into()], "gammaln")
-        .map(CallSiteValue::try_as_basic_value)
-        .map(|v| v.map_left(BasicValueEnum::into_float_value))
-        .map(Either::unwrap_left)
-        .unwrap()
-}
-
-/// Generates a call to `j0` in IR. Returns an `f64` representing the result.
-pub fn call_j0<'ctx>(ctx: &CodeGenContext<'ctx, '_>, v: FloatValue<'ctx>) -> FloatValue<'ctx> {
-    let llvm_f64 = ctx.ctx.f64_type();
-
-    let intrinsic_fn = ctx.module.get_function("__nac3_j0").unwrap_or_else(|| {
-        let fn_type = llvm_f64.fn_type(&[llvm_f64.into()], false);
-        ctx.module.add_function("__nac3_j0", fn_type, None)
-    });
-
-    ctx.builder
-        .build_call(intrinsic_fn, &[v.into()], "j0")
-        .map(CallSiteValue::try_as_basic_value)
-        .map(|v| v.map_left(BasicValueEnum::into_float_value))
-        .map(Either::unwrap_left)
-        .unwrap()
-}
-
-/// Generates a call to `__nac3_ndarray_calc_size`. Returns an [`IntValue`] representing the
-/// calculated total size.
-///
-/// * `dims` - An [`ArrayLikeIndexer`] containing the size of each dimension.
-/// * `range` - The dimension index to begin and end (exclusively) calculating the dimensions for,
-/// or [`None`] if starting from the first dimension and ending at the last dimension respectively.
-pub fn call_ndarray_calc_size<'ctx, G, Dims>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    dims: &Dims,
-    (begin, end): (Option<IntValue<'ctx>>, Option<IntValue<'ctx>>),
-) -> IntValue<'ctx>
-where
-    G: CodeGenerator + ?Sized,
-    Dims: ArrayLikeIndexer<'ctx>,
-{
-    let llvm_usize = generator.get_size_type(ctx.ctx);
-    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
-
-    let ndarray_calc_size_fn_name = match llvm_usize.get_bit_width() {
-        32 => "__nac3_ndarray_calc_size",
-        64 => "__nac3_ndarray_calc_size64",
-        bw => unreachable!("Unsupported size type bit width: {}", bw),
-    };
-    let ndarray_calc_size_fn_t = llvm_usize.fn_type(
-        &[llvm_pusize.into(), llvm_usize.into(), llvm_usize.into(), llvm_usize.into()],
-        false,
-    );
-    let ndarray_calc_size_fn =
-        ctx.module.get_function(ndarray_calc_size_fn_name).unwrap_or_else(|| {
-            ctx.module.add_function(ndarray_calc_size_fn_name, ndarray_calc_size_fn_t, None)
-        });
-
-    let begin = begin.unwrap_or_else(|| llvm_usize.const_zero());
-    let end = end.unwrap_or_else(|| dims.size(ctx, generator));
-    ctx.builder
-        .build_call(
-            ndarray_calc_size_fn,
-            &[
-                dims.base_ptr(ctx, generator).into(),
-                dims.size(ctx, generator).into(),
-                begin.into(),
-                end.into(),
-            ],
-            "",
-        )
-        .map(CallSiteValue::try_as_basic_value)
-        .map(|v| v.map_left(BasicValueEnum::into_int_value))
-        .map(Either::unwrap_left)
-        .unwrap()
-}
-
-/// Generates a call to `__nac3_ndarray_calc_nd_indices`. Returns a [`TypeArrayLikeAdpater`]
-/// containing `i32` indices of the flattened index.
-///
-/// * `index` - The index to compute the multidimensional index for.
-/// * `ndarray` - LLVM pointer to the `NDArray`. This value must be the LLVM representation of an
-/// `NDArray`.
-pub fn call_ndarray_calc_nd_indices<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &G,
-    ctx: &mut CodeGenContext<'ctx, '_>,
-    index: IntValue<'ctx>,
-    ndarray: NDArrayValue<'ctx>,
-) -> TypedArrayLikeAdapter<'ctx, IntValue<'ctx>> {
-    let llvm_void = ctx.ctx.void_type();
-    let llvm_i32 = ctx.ctx.i32_type();
-    let llvm_usize = generator.get_size_type(ctx.ctx);
-    let llvm_pi32 = llvm_i32.ptr_type(AddressSpace::default());
-    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
-
-    let ndarray_calc_nd_indices_fn_name = match llvm_usize.get_bit_width() {
-        32 => "__nac3_ndarray_calc_nd_indices",
-        64 => "__nac3_ndarray_calc_nd_indices64",
-        bw => unreachable!("Unsupported size type bit width: {}", bw),
-    };
-    let ndarray_calc_nd_indices_fn =
-        ctx.module.get_function(ndarray_calc_nd_indices_fn_name).unwrap_or_else(|| {
-            let fn_type = llvm_void.fn_type(
-                &[llvm_usize.into(), llvm_pusize.into(), llvm_usize.into(), llvm_pi32.into()],
-                false,
-            );
-
-            ctx.module.add_function(ndarray_calc_nd_indices_fn_name, fn_type, None)
-        });
-
-    let ndarray_num_dims = ndarray.load_ndims(ctx);
-    let ndarray_dims = ndarray.dim_sizes();
-
-    let indices = ctx.builder.build_array_alloca(llvm_i32, ndarray_num_dims, "").unwrap();
-
-    ctx.builder
-        .build_call(
-            ndarray_calc_nd_indices_fn,
-            &[
-                index.into(),
-                ndarray_dims.base_ptr(ctx, generator).into(),
-                ndarray_num_dims.into(),
-                indices.into(),
-            ],
-            "",
-        )
-        .unwrap();
-
-    TypedArrayLikeAdapter::from(
-        ArraySliceValue::from_ptr_val(indices, ndarray_num_dims, None),
-        Box::new(|_, v| v.into_int_value()),
-        Box::new(|_, v| v.into()),
-    )
-}
-
-fn call_ndarray_flatten_index_impl<'ctx, G, Indices>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    ndarray: NDArrayValue<'ctx>,
-    indices: &Indices,
-) -> IntValue<'ctx>
-where
-    G: CodeGenerator + ?Sized,
-    Indices: ArrayLikeIndexer<'ctx>,
-{
-    let llvm_i32 = ctx.ctx.i32_type();
-    let llvm_usize = generator.get_size_type(ctx.ctx);
-
-    let llvm_pi32 = llvm_i32.ptr_type(AddressSpace::default());
-    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
-
-    debug_assert_eq!(
-        IntType::try_from(indices.element_type(ctx, generator))
-            .map(IntType::get_bit_width)
-            .unwrap_or_default(),
-        llvm_i32.get_bit_width(),
-        "Expected i32 value for argument `indices` to `call_ndarray_flatten_index_impl`"
-    );
-    debug_assert_eq!(
-        indices.size(ctx, generator).get_type().get_bit_width(),
-        llvm_usize.get_bit_width(),
-        "Expected usize integer value for argument `indices_size` to `call_ndarray_flatten_index_impl`"
-    );
-
-    let ndarray_flatten_index_fn_name = match llvm_usize.get_bit_width() {
-        32 => "__nac3_ndarray_flatten_index",
-        64 => "__nac3_ndarray_flatten_index64",
-        bw => unreachable!("Unsupported size type bit width: {}", bw),
-    };
-    let ndarray_flatten_index_fn =
-        ctx.module.get_function(ndarray_flatten_index_fn_name).unwrap_or_else(|| {
-            let fn_type = llvm_usize.fn_type(
-                &[llvm_pusize.into(), llvm_usize.into(), llvm_pi32.into(), llvm_usize.into()],
-                false,
-            );
-
-            ctx.module.add_function(ndarray_flatten_index_fn_name, fn_type, None)
-        });
-
-    let ndarray_num_dims = ndarray.load_ndims(ctx);
-    let ndarray_dims = ndarray.dim_sizes();
-
-    let index = ctx
-        .builder
-        .build_call(
-            ndarray_flatten_index_fn,
-            &[
-                ndarray_dims.base_ptr(ctx, generator).into(),
-                ndarray_num_dims.into(),
-                indices.base_ptr(ctx, generator).into(),
-                indices.size(ctx, generator).into(),
-            ],
-            "",
-        )
-        .map(CallSiteValue::try_as_basic_value)
-        .map(|v| v.map_left(BasicValueEnum::into_int_value))
-        .map(Either::unwrap_left)
-        .unwrap();
-
-    index
-}
-
-/// Generates a call to `__nac3_ndarray_flatten_index`. Returns the flattened index for the
-/// multidimensional index.
-///
-/// * `ndarray` - LLVM pointer to the `NDArray`. This value must be the LLVM representation of an
-/// `NDArray`.
-/// * `indices` - The multidimensional index to compute the flattened index for.
-pub fn call_ndarray_flatten_index<'ctx, G, Index>(
-    generator: &mut G,
-    ctx: &mut CodeGenContext<'ctx, '_>,
-    ndarray: NDArrayValue<'ctx>,
-    indices: &Index,
-) -> IntValue<'ctx>
-where
-    G: CodeGenerator + ?Sized,
-    Index: ArrayLikeIndexer<'ctx>,
-{
-    call_ndarray_flatten_index_impl(generator, ctx, ndarray, indices)
-}
-
-/// Generates a call to `__nac3_ndarray_calc_broadcast`. Returns a tuple containing the number of
-/// dimension and size of each dimension of the resultant `ndarray`.
-pub fn call_ndarray_calc_broadcast<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &mut G,
-    ctx: &mut CodeGenContext<'ctx, '_>,
-    lhs: NDArrayValue<'ctx>,
-    rhs: NDArrayValue<'ctx>,
-) -> TypedArrayLikeAdapter<'ctx, IntValue<'ctx>> {
-    let llvm_usize = generator.get_size_type(ctx.ctx);
-    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
-
-    let ndarray_calc_broadcast_fn_name = match llvm_usize.get_bit_width() {
-        32 => "__nac3_ndarray_calc_broadcast",
-        64 => "__nac3_ndarray_calc_broadcast64",
-        bw => unreachable!("Unsupported size type bit width: {}", bw),
-    };
-    let ndarray_calc_broadcast_fn =
-        ctx.module.get_function(ndarray_calc_broadcast_fn_name).unwrap_or_else(|| {
-            let fn_type = llvm_usize.fn_type(
-                &[
-                    llvm_pusize.into(),
-                    llvm_usize.into(),
-                    llvm_pusize.into(),
-                    llvm_usize.into(),
-                    llvm_pusize.into(),
-                ],
-                false,
-            );
-
-            ctx.module.add_function(ndarray_calc_broadcast_fn_name, fn_type, None)
-        });
-
-    let lhs_ndims = lhs.load_ndims(ctx);
-    let rhs_ndims = rhs.load_ndims(ctx);
-    let min_ndims = llvm_intrinsics::call_int_umin(ctx, lhs_ndims, rhs_ndims, None);
-
-    gen_for_callback_incrementing(
-        generator,
-        ctx,
-        llvm_usize.const_zero(),
-        (min_ndims, false),
-        |generator, ctx, _, idx| {
-            let idx = ctx.builder.build_int_sub(min_ndims, idx, "").unwrap();
-            let (lhs_dim_sz, rhs_dim_sz) = unsafe {
-                (
-                    lhs.dim_sizes().get_typed_unchecked(ctx, generator, &idx, None),
-                    rhs.dim_sizes().get_typed_unchecked(ctx, generator, &idx, None),
-                )
-            };
-
-            let llvm_usize_const_one = llvm_usize.const_int(1, false);
-            let lhs_eqz = ctx
-                .builder
-                .build_int_compare(IntPredicate::EQ, lhs_dim_sz, llvm_usize_const_one, "")
-                .unwrap();
-            let rhs_eqz = ctx
-                .builder
-                .build_int_compare(IntPredicate::EQ, rhs_dim_sz, llvm_usize_const_one, "")
-                .unwrap();
-            let lhs_or_rhs_eqz = ctx.builder.build_or(lhs_eqz, rhs_eqz, "").unwrap();
-
-            let lhs_eq_rhs = ctx
-                .builder
-                .build_int_compare(IntPredicate::EQ, lhs_dim_sz, rhs_dim_sz, "")
-                .unwrap();
-
-            let is_compatible = ctx.builder.build_or(lhs_or_rhs_eqz, lhs_eq_rhs, "").unwrap();
-
-            ctx.make_assert(
-                generator,
-                is_compatible,
-                "0:ValueError",
-                "operands could not be broadcast together",
-                [None, None, None],
-                ctx.current_loc,
-            );
-
-            Ok(())
-        },
-        llvm_usize.const_int(1, false),
-    )
-    .unwrap();
-
-    let max_ndims = llvm_intrinsics::call_int_umax(ctx, lhs_ndims, rhs_ndims, None);
-    let lhs_dims = lhs.dim_sizes().base_ptr(ctx, generator);
-    let lhs_ndims = lhs.load_ndims(ctx);
-    let rhs_dims = rhs.dim_sizes().base_ptr(ctx, generator);
-    let rhs_ndims = rhs.load_ndims(ctx);
-    let out_dims = ctx.builder.build_array_alloca(llvm_usize, max_ndims, "").unwrap();
-    let out_dims = ArraySliceValue::from_ptr_val(out_dims, max_ndims, None);
-
-    ctx.builder
-        .build_call(
-            ndarray_calc_broadcast_fn,
-            &[
-                lhs_dims.into(),
-                lhs_ndims.into(),
-                rhs_dims.into(),
-                rhs_ndims.into(),
-                out_dims.base_ptr(ctx, generator).into(),
-            ],
-            "",
-        )
-        .unwrap();
-
-    TypedArrayLikeAdapter::from(
-        out_dims,
-        Box::new(|_, v| v.into_int_value()),
-        Box::new(|_, v| v.into()),
-    )
-}
-
-/// Generates a call to `__nac3_ndarray_calc_broadcast_idx`. Returns an [`ArrayAllocaValue`]
-/// containing the indices used for accessing `array` corresponding to the index of the broadcasted
-/// array `broadcast_idx`.
-pub fn call_ndarray_calc_broadcast_index<
-    'ctx,
-    G: CodeGenerator + ?Sized,
-    BroadcastIdx: UntypedArrayLikeAccessor<'ctx>,
->(
-    generator: &mut G,
-    ctx: &mut CodeGenContext<'ctx, '_>,
-    array: NDArrayValue<'ctx>,
-    broadcast_idx: &BroadcastIdx,
-) -> TypedArrayLikeAdapter<'ctx, IntValue<'ctx>> {
-    let llvm_i32 = ctx.ctx.i32_type();
-    let llvm_usize = generator.get_size_type(ctx.ctx);
-    let llvm_pi32 = llvm_i32.ptr_type(AddressSpace::default());
-    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
-
-    let ndarray_calc_broadcast_fn_name = match llvm_usize.get_bit_width() {
-        32 => "__nac3_ndarray_calc_broadcast_idx",
-        64 => "__nac3_ndarray_calc_broadcast_idx64",
-        bw => unreachable!("Unsupported size type bit width: {}", bw),
-    };
-    let ndarray_calc_broadcast_fn =
-        ctx.module.get_function(ndarray_calc_broadcast_fn_name).unwrap_or_else(|| {
-            let fn_type = llvm_usize.fn_type(
-                &[llvm_pusize.into(), llvm_usize.into(), llvm_pi32.into(), llvm_pi32.into()],
-                false,
-            );
-
-            ctx.module.add_function(ndarray_calc_broadcast_fn_name, fn_type, None)
-        });
-
-    let broadcast_size = broadcast_idx.size(ctx, generator);
-    let out_idx = ctx.builder.build_array_alloca(llvm_i32, broadcast_size, "").unwrap();
-
-    let array_dims = array.dim_sizes().base_ptr(ctx, generator);
-    let array_ndims = array.load_ndims(ctx);
-    let broadcast_idx_ptr = unsafe {
-        broadcast_idx.ptr_offset_unchecked(ctx, generator, &llvm_usize.const_zero(), None)
-    };
-
-    ctx.builder
-        .build_call(
-            ndarray_calc_broadcast_fn,
-            &[array_dims.into(), array_ndims.into(), broadcast_idx_ptr.into(), out_idx.into()],
-            "",
-        )
-        .unwrap();
-
-    TypedArrayLikeAdapter::from(
-        ArraySliceValue::from_ptr_val(out_idx, broadcast_size, None),
-        Box::new(|_, v| v.into_int_value()),
-        Box::new(|_, v| v.into()),
-    )
-}
-
-fn get_size_variant<'ctx>(ty: IntType<'ctx>) -> SizeVariant {
-    match ty.get_bit_width() {
-        32 => SizeVariant::Bits32,
-        64 => SizeVariant::Bits64,
-        _ => unreachable!("Unsupported int type bit width {}", ty.get_bit_width()),
-    }
-}
-
-fn get_size_type_dependent_function<'ctx, BuildFuncTypeFn>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    size_type: IntType<'ctx>,
-    base_name: &str,
-    build_func_type: BuildFuncTypeFn,
-) -> FunctionValue<'ctx>
-where
-    BuildFuncTypeFn: Fn() -> FunctionType<'ctx>,
-{
-    let mut fn_name = base_name.to_owned();
-    match get_size_variant(size_type) {
-        SizeVariant::Bits32 => {
-            // The original fn_name is the correct function name
-        }
-        SizeVariant::Bits64 => {
-            // Append "64" at the end, this is the naming convention for 64-bit
-            fn_name.push_str("64");
-        }
-    }
-
-    // Get (or declare then get if does not exist) the corresponding function
-    ctx.module.get_function(&fn_name).unwrap_or_else(|| {
-        let fn_type = build_func_type();
-        ctx.module.add_function(&fn_name, fn_type, None)
-    })
-}
-
-fn get_ndarray_struct_ptr<'ctx>(ctx: &'ctx Context, size_type: IntType<'ctx>) -> PointerType<'ctx> {
-    let i8_type = ctx.i8_type();
-
-    let ndarray_ty = NpArrayType { size_type, elem_type: i8_type.as_basic_type_enum() };
-    let struct_ty = ndarray_ty.fields().whole_struct.as_struct_type(ctx);
-    struct_ty.ptr_type(AddressSpace::default())
-}
-
-pub fn call_nac3_ndarray_size<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    ndarray: NpArrayValue<'ctx>,
-) -> IntValue<'ctx> {
-    let size_type = ndarray.ty.size_type;
-    let function = get_size_type_dependent_function(ctx, size_type, "__nac3_ndarray_size", || {
-        size_type.fn_type(&[get_ndarray_struct_ptr(ctx.ctx, size_type).into()], false)
-    });
-
-    ctx.builder
-        .build_call(function, &[ndarray.ptr.into()], "size")
-        .unwrap()
-        .try_as_basic_value()
-        .unwrap_left()
-        .into_int_value()
-}

nac3core/src/codegen/irrt/ndarray.rs

@@ -0,0 +1,384 @@
+use inkwell::{
+    types::IntType,
+    values::{BasicValueEnum, CallSiteValue, IntValue},
+    AddressSpace, IntPredicate,
+};
+use itertools::Either;
+
+use crate::codegen::{
+    llvm_intrinsics,
+    macros::codegen_unreachable,
+    stmt::gen_for_callback_incrementing,
+    values::{
+        ArrayLikeIndexer, ArrayLikeValue, ArraySliceValue, NDArrayValue, TypedArrayLikeAccessor,
+        TypedArrayLikeAdapter, UntypedArrayLikeAccessor,
+    },
+    CodeGenContext, CodeGenerator,
+};
+
+/// Generates a call to `__nac3_ndarray_calc_size`. Returns an [`IntValue`] representing the
+/// calculated total size.
+///
+/// * `dims` - An [`ArrayLikeIndexer`] containing the size of each dimension.
+/// * `range` - The dimension index to begin and end (exclusively) calculating the dimensions for,
+///   or [`None`] if starting from the first dimension and ending at the last dimension
+///   respectively.
+pub fn call_ndarray_calc_size<'ctx, G, Dims>(
+    generator: &G,
+    ctx: &CodeGenContext<'ctx, '_>,
+    dims: &Dims,
+    (begin, end): (Option<IntValue<'ctx>>, Option<IntValue<'ctx>>),
+) -> IntValue<'ctx>
+where
+    G: CodeGenerator + ?Sized,
+    Dims: ArrayLikeIndexer<'ctx>,
+{
+    let llvm_usize = generator.get_size_type(ctx.ctx);
+    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
+
+    let ndarray_calc_size_fn_name = match llvm_usize.get_bit_width() {
+        32 => "__nac3_ndarray_calc_size",
+        64 => "__nac3_ndarray_calc_size64",
+        bw => codegen_unreachable!(ctx, "Unsupported size type bit width: {}", bw),
+    };
+    let ndarray_calc_size_fn_t = llvm_usize.fn_type(
+        &[llvm_pusize.into(), llvm_usize.into(), llvm_usize.into(), llvm_usize.into()],
+        false,
+    );
+    let ndarray_calc_size_fn =
+        ctx.module.get_function(ndarray_calc_size_fn_name).unwrap_or_else(|| {
+            ctx.module.add_function(ndarray_calc_size_fn_name, ndarray_calc_size_fn_t, None)
+        });
+
+    let begin = begin.unwrap_or_else(|| llvm_usize.const_zero());
+    let end = end.unwrap_or_else(|| dims.size(ctx, generator));
+    ctx.builder
+        .build_call(
+            ndarray_calc_size_fn,
+            &[
+                dims.base_ptr(ctx, generator).into(),
+                dims.size(ctx, generator).into(),
+                begin.into(),
+                end.into(),
+            ],
+            "",
+        )
+        .map(CallSiteValue::try_as_basic_value)
+        .map(|v| v.map_left(BasicValueEnum::into_int_value))
+        .map(Either::unwrap_left)
+        .unwrap()
+}
+
+/// Generates a call to `__nac3_ndarray_calc_nd_indices`. Returns a [`TypeArrayLikeAdpater`]
+/// containing `i32` indices of the flattened index.
+///
+/// * `index` - The index to compute the multidimensional index for.
+/// * `ndarray` - LLVM pointer to the `NDArray`. This value must be the LLVM representation of an
+///   `NDArray`.
+pub fn call_ndarray_calc_nd_indices<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    index: IntValue<'ctx>,
+    ndarray: NDArrayValue<'ctx>,
+) -> TypedArrayLikeAdapter<'ctx, IntValue<'ctx>> {
+    let llvm_void = ctx.ctx.void_type();
+    let llvm_i32 = ctx.ctx.i32_type();
+    let llvm_usize = generator.get_size_type(ctx.ctx);
+    let llvm_pi32 = llvm_i32.ptr_type(AddressSpace::default());
+    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
+
+    let ndarray_calc_nd_indices_fn_name = match llvm_usize.get_bit_width() {
+        32 => "__nac3_ndarray_calc_nd_indices",
+        64 => "__nac3_ndarray_calc_nd_indices64",
+        bw => codegen_unreachable!(ctx, "Unsupported size type bit width: {}", bw),
+    };
+    let ndarray_calc_nd_indices_fn =
+        ctx.module.get_function(ndarray_calc_nd_indices_fn_name).unwrap_or_else(|| {
+            let fn_type = llvm_void.fn_type(
+                &[llvm_usize.into(), llvm_pusize.into(), llvm_usize.into(), llvm_pi32.into()],
+                false,
+            );
+
+            ctx.module.add_function(ndarray_calc_nd_indices_fn_name, fn_type, None)
+        });
+
+    let ndarray_num_dims = ndarray.load_ndims(ctx);
+    let ndarray_dims = ndarray.shape();
+
+    let indices = ctx.builder.build_array_alloca(llvm_i32, ndarray_num_dims, "").unwrap();
+
+    ctx.builder
+        .build_call(
+            ndarray_calc_nd_indices_fn,
+            &[
+                index.into(),
+                ndarray_dims.base_ptr(ctx, generator).into(),
+                ndarray_num_dims.into(),
+                indices.into(),
+            ],
+            "",
+        )
+        .unwrap();
+
+    TypedArrayLikeAdapter::from(
+        ArraySliceValue::from_ptr_val(indices, ndarray_num_dims, None),
+        Box::new(|_, v| v.into_int_value()),
+        Box::new(|_, v| v.into()),
+    )
+}
+
+fn call_ndarray_flatten_index_impl<'ctx, G, Indices>(
+    generator: &G,
+    ctx: &CodeGenContext<'ctx, '_>,
+    ndarray: NDArrayValue<'ctx>,
+    indices: &Indices,
+) -> IntValue<'ctx>
+where
+    G: CodeGenerator + ?Sized,
+    Indices: ArrayLikeIndexer<'ctx>,
+{
+    let llvm_i32 = ctx.ctx.i32_type();
+    let llvm_usize = generator.get_size_type(ctx.ctx);
+
+    let llvm_pi32 = llvm_i32.ptr_type(AddressSpace::default());
+    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
+
+    debug_assert_eq!(
+        IntType::try_from(indices.element_type(ctx, generator))
+            .map(IntType::get_bit_width)
+            .unwrap_or_default(),
+        llvm_i32.get_bit_width(),
+        "Expected i32 value for argument `indices` to `call_ndarray_flatten_index_impl`"
+    );
+    debug_assert_eq!(
+        indices.size(ctx, generator).get_type().get_bit_width(),
+        llvm_usize.get_bit_width(),
+        "Expected usize integer value for argument `indices_size` to `call_ndarray_flatten_index_impl`"
+    );
+
+    let ndarray_flatten_index_fn_name = match llvm_usize.get_bit_width() {
+        32 => "__nac3_ndarray_flatten_index",
+        64 => "__nac3_ndarray_flatten_index64",
+        bw => codegen_unreachable!(ctx, "Unsupported size type bit width: {}", bw),
+    };
+    let ndarray_flatten_index_fn =
+        ctx.module.get_function(ndarray_flatten_index_fn_name).unwrap_or_else(|| {
+            let fn_type = llvm_usize.fn_type(
+                &[llvm_pusize.into(), llvm_usize.into(), llvm_pi32.into(), llvm_usize.into()],
+                false,
+            );
+
+            ctx.module.add_function(ndarray_flatten_index_fn_name, fn_type, None)
+        });
+
+    let ndarray_num_dims = ndarray.load_ndims(ctx);
+    let ndarray_dims = ndarray.shape();
+
+    let index = ctx
+        .builder
+        .build_call(
+            ndarray_flatten_index_fn,
+            &[
+                ndarray_dims.base_ptr(ctx, generator).into(),
+                ndarray_num_dims.into(),
+                indices.base_ptr(ctx, generator).into(),
+                indices.size(ctx, generator).into(),
+            ],
+            "",
+        )
+        .map(CallSiteValue::try_as_basic_value)
+        .map(|v| v.map_left(BasicValueEnum::into_int_value))
+        .map(Either::unwrap_left)
+        .unwrap();
+
+    index
+}
+
+/// Generates a call to `__nac3_ndarray_flatten_index`. Returns the flattened index for the
+/// multidimensional index.
+///
+/// * `ndarray` - LLVM pointer to the `NDArray`. This value must be the LLVM representation of an
+///   `NDArray`.
+/// * `indices` - The multidimensional index to compute the flattened index for.
+pub fn call_ndarray_flatten_index<'ctx, G, Index>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    ndarray: NDArrayValue<'ctx>,
+    indices: &Index,
+) -> IntValue<'ctx>
+where
+    G: CodeGenerator + ?Sized,
+    Index: ArrayLikeIndexer<'ctx>,
+{
+    call_ndarray_flatten_index_impl(generator, ctx, ndarray, indices)
+}
+
+/// Generates a call to `__nac3_ndarray_calc_broadcast`. Returns a tuple containing the number of
+/// dimension and size of each dimension of the resultant `ndarray`.
+pub fn call_ndarray_calc_broadcast<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    lhs: NDArrayValue<'ctx>,
+    rhs: NDArrayValue<'ctx>,
+) -> TypedArrayLikeAdapter<'ctx, IntValue<'ctx>> {
+    let llvm_usize = generator.get_size_type(ctx.ctx);
+    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
+
+    let ndarray_calc_broadcast_fn_name = match llvm_usize.get_bit_width() {
+        32 => "__nac3_ndarray_calc_broadcast",
+        64 => "__nac3_ndarray_calc_broadcast64",
+        bw => codegen_unreachable!(ctx, "Unsupported size type bit width: {}", bw),
+    };
+    let ndarray_calc_broadcast_fn =
+        ctx.module.get_function(ndarray_calc_broadcast_fn_name).unwrap_or_else(|| {
+            let fn_type = llvm_usize.fn_type(
+                &[
+                    llvm_pusize.into(),
+                    llvm_usize.into(),
+                    llvm_pusize.into(),
+                    llvm_usize.into(),
+                    llvm_pusize.into(),
+                ],
+                false,
+            );
+
+            ctx.module.add_function(ndarray_calc_broadcast_fn_name, fn_type, None)
+        });
+
+    let lhs_ndims = lhs.load_ndims(ctx);
+    let rhs_ndims = rhs.load_ndims(ctx);
+    let min_ndims = llvm_intrinsics::call_int_umin(ctx, lhs_ndims, rhs_ndims, None);
+
+    gen_for_callback_incrementing(
+        generator,
+        ctx,
+        None,
+        llvm_usize.const_zero(),
+        (min_ndims, false),
+        |generator, ctx, _, idx| {
+            let idx = ctx.builder.build_int_sub(min_ndims, idx, "").unwrap();
+            let (lhs_dim_sz, rhs_dim_sz) = unsafe {
+                (
+                    lhs.shape().get_typed_unchecked(ctx, generator, &idx, None),
+                    rhs.shape().get_typed_unchecked(ctx, generator, &idx, None),
+                )
+            };
+
+            let llvm_usize_const_one = llvm_usize.const_int(1, false);
+            let lhs_eqz = ctx
+                .builder
+                .build_int_compare(IntPredicate::EQ, lhs_dim_sz, llvm_usize_const_one, "")
+                .unwrap();
+            let rhs_eqz = ctx
+                .builder
+                .build_int_compare(IntPredicate::EQ, rhs_dim_sz, llvm_usize_const_one, "")
+                .unwrap();
+            let lhs_or_rhs_eqz = ctx.builder.build_or(lhs_eqz, rhs_eqz, "").unwrap();
+
+            let lhs_eq_rhs = ctx
+                .builder
+                .build_int_compare(IntPredicate::EQ, lhs_dim_sz, rhs_dim_sz, "")
+                .unwrap();
+
+            let is_compatible = ctx.builder.build_or(lhs_or_rhs_eqz, lhs_eq_rhs, "").unwrap();
+
+            ctx.make_assert(
+                generator,
+                is_compatible,
+                "0:ValueError",
+                "operands could not be broadcast together",
+                [None, None, None],
+                ctx.current_loc,
+            );
+
+            Ok(())
+        },
+        llvm_usize.const_int(1, false),
+    )
+    .unwrap();
+
+    let max_ndims = llvm_intrinsics::call_int_umax(ctx, lhs_ndims, rhs_ndims, None);
+    let lhs_dims = lhs.shape().base_ptr(ctx, generator);
+    let lhs_ndims = lhs.load_ndims(ctx);
+    let rhs_dims = rhs.shape().base_ptr(ctx, generator);
+    let rhs_ndims = rhs.load_ndims(ctx);
+    let out_dims = ctx.builder.build_array_alloca(llvm_usize, max_ndims, "").unwrap();
+    let out_dims = ArraySliceValue::from_ptr_val(out_dims, max_ndims, None);
+
+    ctx.builder
+        .build_call(
+            ndarray_calc_broadcast_fn,
+            &[
+                lhs_dims.into(),
+                lhs_ndims.into(),
+                rhs_dims.into(),
+                rhs_ndims.into(),
+                out_dims.base_ptr(ctx, generator).into(),
+            ],
+            "",
+        )
+        .unwrap();
+
+    TypedArrayLikeAdapter::from(
+        out_dims,
+        Box::new(|_, v| v.into_int_value()),
+        Box::new(|_, v| v.into()),
+    )
+}
+
+/// Generates a call to `__nac3_ndarray_calc_broadcast_idx`. Returns an [`ArrayAllocaValue`]
+/// containing the indices used for accessing `array` corresponding to the index of the broadcasted
+/// array `broadcast_idx`.
+pub fn call_ndarray_calc_broadcast_index<
+    'ctx,
+    G: CodeGenerator + ?Sized,
+    BroadcastIdx: UntypedArrayLikeAccessor<'ctx>,
+>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    array: NDArrayValue<'ctx>,
+    broadcast_idx: &BroadcastIdx,
+) -> TypedArrayLikeAdapter<'ctx, IntValue<'ctx>> {
+    let llvm_i32 = ctx.ctx.i32_type();
+    let llvm_usize = generator.get_size_type(ctx.ctx);
+    let llvm_pi32 = llvm_i32.ptr_type(AddressSpace::default());
+    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
+
+    let ndarray_calc_broadcast_fn_name = match llvm_usize.get_bit_width() {
+        32 => "__nac3_ndarray_calc_broadcast_idx",
+        64 => "__nac3_ndarray_calc_broadcast_idx64",
+        bw => codegen_unreachable!(ctx, "Unsupported size type bit width: {}", bw),
+    };
+    let ndarray_calc_broadcast_fn =
+        ctx.module.get_function(ndarray_calc_broadcast_fn_name).unwrap_or_else(|| {
+            let fn_type = llvm_usize.fn_type(
+                &[llvm_pusize.into(), llvm_usize.into(), llvm_pi32.into(), llvm_pi32.into()],
+                false,
+            );
+
+            ctx.module.add_function(ndarray_calc_broadcast_fn_name, fn_type, None)
+        });
+
+    let broadcast_size = broadcast_idx.size(ctx, generator);
+    let out_idx = ctx.builder.build_array_alloca(llvm_i32, broadcast_size, "").unwrap();
+
+    let array_dims = array.shape().base_ptr(ctx, generator);
+    let array_ndims = array.load_ndims(ctx);
+    let broadcast_idx_ptr = unsafe {
+        broadcast_idx.ptr_offset_unchecked(ctx, generator, &llvm_usize.const_zero(), None)
+    };
+
+    ctx.builder
+        .build_call(
+            ndarray_calc_broadcast_fn,
+            &[array_dims.into(), array_ndims.into(), broadcast_idx_ptr.into(), out_idx.into()],
+            "",
+        )
+        .unwrap();
+
+    TypedArrayLikeAdapter::from(
+        ArraySliceValue::from_ptr_val(out_idx, broadcast_size, None),
+        Box::new(|_, v| v.into_int_value()),
+        Box::new(|_, v| v.into()),
+    )
+}

nac3core/src/codegen/irrt/slice.rs

@@ -0,0 +1,76 @@
+use inkwell::{
+    values::{BasicValueEnum, CallSiteValue, IntValue},
+    IntPredicate,
+};
+use itertools::Either;
+use nac3parser::ast::Expr;
+
+use crate::{
+    codegen::{CodeGenContext, CodeGenerator},
+    typecheck::typedef::Type,
+};
+
+/// this function allows index out of range, since python
+/// allows index out of range in slice (`a = [1,2,3]; a[1:10] == [2,3]`).
+pub fn handle_slice_index_bound<'ctx, G: CodeGenerator>(
+    i: &Expr<Option<Type>>,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    generator: &mut G,
+    length: IntValue<'ctx>,
+) -> Result<Option<IntValue<'ctx>>, String> {
+    const SYMBOL: &str = "__nac3_slice_index_bound";
+    let func = ctx.module.get_function(SYMBOL).unwrap_or_else(|| {
+        let i32_t = ctx.ctx.i32_type();
+        let fn_t = i32_t.fn_type(&[i32_t.into(), i32_t.into()], false);
+        ctx.module.add_function(SYMBOL, fn_t, None)
+    });
+
+    let i = if let Some(v) = generator.gen_expr(ctx, i)? {
+        v.to_basic_value_enum(ctx, generator, i.custom.unwrap())?
+    } else {
+        return Ok(None);
+    };
+    Ok(Some(
+        ctx.builder
+            .build_call(func, &[i.into(), length.into()], "bounded_ind")
+            .map(CallSiteValue::try_as_basic_value)
+            .map(|v| v.map_left(BasicValueEnum::into_int_value))
+            .map(Either::unwrap_left)
+            .unwrap(),
+    ))
+}
+
+pub fn calculate_len_for_slice_range<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    start: IntValue<'ctx>,
+    end: IntValue<'ctx>,
+    step: IntValue<'ctx>,
+) -> IntValue<'ctx> {
+    const SYMBOL: &str = "__nac3_range_slice_len";
+    let len_func = ctx.module.get_function(SYMBOL).unwrap_or_else(|| {
+        let i32_t = ctx.ctx.i32_type();
+        let fn_t = i32_t.fn_type(&[i32_t.into(), i32_t.into(), i32_t.into()], false);
+        ctx.module.add_function(SYMBOL, fn_t, None)
+    });
+
+    // assert step != 0, throw exception if not
+    let not_zero = ctx
+        .builder
+        .build_int_compare(IntPredicate::NE, step, step.get_type().const_zero(), "range_step_ne")
+        .unwrap();
+    ctx.make_assert(
+        generator,
+        not_zero,
+        "0:ValueError",
+        "step must not be zero",
+        [None, None, None],
+        ctx.current_loc,
+    );
+    ctx.builder
+        .build_call(len_func, &[start.into(), end.into(), step.into()], "calc_len")
+        .map(CallSiteValue::try_as_basic_value)
+        .map(|v| v.map_left(BasicValueEnum::into_int_value))
+        .map(Either::unwrap_left)
+        .unwrap()
+}

nac3core/src/codegen/irrt/test.rs

@@ -1,26 +0,0 @@
-#[cfg(test)]
-mod tests {
-    use std::{path::Path, process::Command};
-
-    #[test]
-    fn run_irrt_test() {
-        assert!(
-            cfg!(feature = "test"),
-            "Please do `cargo test -F test` to compile `irrt_test.out` and run test"
-        );
-
-        let irrt_test_out_path = Path::new(concat!(env!("OUT_DIR"), "/irrt_test.out"));
-        let output = Command::new(irrt_test_out_path.to_str().unwrap()).output().unwrap();
-
-        if !output.status.success() {
-            eprintln!("irrt_test failed with status {}:", output.status);
-            eprintln!("====== stdout ======");
-            eprintln!("{}", String::from_utf8(output.stdout).unwrap());
-            eprintln!("====== stderr ======");
-            eprintln!("{}", String::from_utf8(output.stderr).unwrap());
-            eprintln!("====================");
-
-            panic!("irrt_test failed");
-        }
-    }
-}

nac3core/src/codegen/llvm_intrinsics.rs

@@ -1,12 +1,14 @@
-use crate::codegen::CodeGenContext;
-use inkwell::context::Context;
-use inkwell::intrinsics::Intrinsic;
-use inkwell::types::AnyTypeEnum::IntType;
-use inkwell::types::FloatType;
-use inkwell::values::{BasicValueEnum, CallSiteValue, FloatValue, IntValue, PointerValue};
-use inkwell::AddressSpace;
+use inkwell::{
+    context::Context,
+    intrinsics::Intrinsic,
+    types::{AnyTypeEnum::IntType, FloatType},
+    values::{BasicValueEnum, CallSiteValue, FloatValue, IntValue, PointerValue},
+    AddressSpace,
+};
 use itertools::Either;
 
+use super::CodeGenContext;
+
 /// Returns the string representation for the floating-point type `ft` when used in intrinsic
 /// functions.
 fn get_float_intrinsic_repr(ctx: &Context, ft: FloatType) -> &'static str {
@@ -35,6 +37,40 @@ fn get_float_intrinsic_repr(ctx: &Context, ft: FloatType) -> &'static str {
     unreachable!()
 }
 
+/// Invokes the [`llvm.va_start`](https://llvm.org/docs/LangRef.html#llvm-va-start-intrinsic)
+/// intrinsic.
+pub fn call_va_start<'ctx>(ctx: &CodeGenContext<'ctx, '_>, arglist: PointerValue<'ctx>) {
+    const FN_NAME: &str = "llvm.va_start";
+
+    let intrinsic_fn = ctx.module.get_function(FN_NAME).unwrap_or_else(|| {
+        let llvm_void = ctx.ctx.void_type();
+        let llvm_i8 = ctx.ctx.i8_type();
+        let llvm_p0i8 = llvm_i8.ptr_type(AddressSpace::default());
+        let fn_type = llvm_void.fn_type(&[llvm_p0i8.into()], false);
+
+        ctx.module.add_function(FN_NAME, fn_type, None)
+    });
+
+    ctx.builder.build_call(intrinsic_fn, &[arglist.into()], "").unwrap();
+}
+
+/// Invokes the [`llvm.va_start`](https://llvm.org/docs/LangRef.html#llvm-va-start-intrinsic)
+/// intrinsic.
+pub fn call_va_end<'ctx>(ctx: &CodeGenContext<'ctx, '_>, arglist: PointerValue<'ctx>) {
+    const FN_NAME: &str = "llvm.va_end";
+
+    let intrinsic_fn = ctx.module.get_function(FN_NAME).unwrap_or_else(|| {
+        let llvm_void = ctx.ctx.void_type();
+        let llvm_i8 = ctx.ctx.i8_type();
+        let llvm_p0i8 = llvm_i8.ptr_type(AddressSpace::default());
+        let fn_type = llvm_void.fn_type(&[llvm_p0i8.into()], false);
+
+        ctx.module.add_function(FN_NAME, fn_type, None)
+    });
+
+    ctx.builder.build_call(intrinsic_fn, &[arglist.into()], "").unwrap();
+}
+
 /// Invokes the [`llvm.stacksave`](https://llvm.org/docs/LangRef.html#llvm-stacksave-intrinsic)
 /// intrinsic.
 pub fn call_stacksave<'ctx>(
@@ -149,7 +185,7 @@ pub fn call_memcpy_generic<'ctx>(
         dest
     } else {
         ctx.builder
-            .build_bitcast(dest, llvm_p0i8, "")
+            .build_bit_cast(dest, llvm_p0i8, "")
             .map(BasicValueEnum::into_pointer_value)
             .unwrap()
     };
@@ -157,7 +193,7 @@ pub fn call_memcpy_generic<'ctx>(
         src
     } else {
         ctx.builder
-            .build_bitcast(src, llvm_p0i8, "")
+            .build_bit_cast(src, llvm_p0i8, "")
             .map(BasicValueEnum::into_pointer_value)
             .unwrap()
     };
@@ -171,8 +207,9 @@ pub fn call_memcpy_generic<'ctx>(
 /// * `$ctx:ident`: Reference to the current Code Generation Context
 /// * `$name:ident`: Optional name to be assigned to the llvm build call (Option<&str>)
 /// * `$llvm_name:literal`: Name of underlying llvm intrinsic function
-/// * `$map_fn:ident`: Mapping function to be applied on `BasicValue` (`BasicValue` -> Function Return Type)
-/// Use `BasicValueEnum::into_int_value` for Integer return type and `BasicValueEnum::into_float_value` for Float return type
+/// * `$map_fn:ident`: Mapping function to be applied on `BasicValue` (`BasicValue` -> Function Return Type).
+///   Use `BasicValueEnum::into_int_value` for Integer return type and
+///   `BasicValueEnum::into_float_value` for Float return type
 /// * `$llvm_ty:ident`: Type of first operand
 /// * `,($val:ident)*`: Comma separated list of operands
 macro_rules! generate_llvm_intrinsic_fn_body {
@@ -188,7 +225,7 @@ macro_rules! generate_llvm_intrinsic_fn_body {
 /// Arguments:
 /// * `float/int`: Indicates the return and argument type of the function
 /// * `$fn_name:ident`: The identifier of the rust function to be generated
-/// * `$llvm_name:literal`: Name of underlying llvm intrinsic function
+/// * `$llvm_name:literal`: Name of underlying llvm intrinsic function.
 ///   Omit "llvm." prefix from the function name i.e. use "ceil" instead of "llvm.ceil"
 /// * `$val:ident`: The operand for unary operations
 /// * `$val1:ident`, `$val2:ident`: The operands for binary operations

nac3core/src/codegen/mod.rs

@@ -1,12 +1,12 @@
-use crate::{
-    codegen::classes::{ListType, NDArrayType, ProxyType, RangeType},
-    symbol_resolver::{StaticValue, SymbolResolver},
-    toplevel::{helper::PrimDef, numpy::unpack_ndarray_var_tys, TopLevelContext, TopLevelDef},
-    typecheck::{
-        type_inferencer::{CodeLocation, PrimitiveStore},
-        typedef::{CallId, FuncArg, Type, TypeEnum, Unifier},
+use std::{
+    collections::{HashMap, HashSet},
+    sync::{
+        atomic::{AtomicBool, Ordering},
+        Arc,
     },
+    thread,
 };
+
 use crossbeam::channel::{unbounded, Receiver, Sender};
 use inkwell::{
     attributes::{Attribute, AttributeLoc},
@@ -24,17 +24,23 @@ use inkwell::{
     AddressSpace, IntPredicate, OptimizationLevel,
 };
 use itertools::Itertools;
-use nac3parser::ast::{Location, Stmt, StrRef};
 use parking_lot::{Condvar, Mutex};
-use std::collections::{HashMap, HashSet};
-use std::sync::{
-    atomic::{AtomicBool, Ordering},
-    Arc,
+
+use nac3parser::ast::{Location, Stmt, StrRef};
+
+use crate::{
+    symbol_resolver::{StaticValue, SymbolResolver},
+    toplevel::{helper::PrimDef, numpy::unpack_ndarray_var_tys, TopLevelContext, TopLevelDef},
+    typecheck::{
+        type_inferencer::{CodeLocation, PrimitiveStore},
+        typedef::{CallId, FuncArg, Type, TypeEnum, Unifier},
+    },
 };
-use std::thread;
+use concrete_type::{ConcreteType, ConcreteTypeEnum, ConcreteTypeStore};
+pub use generator::{CodeGenerator, DefaultCodeGenerator};
+use types::{ListType, NDArrayType, ProxyType, RangeType};
 
 pub mod builtin_fns;
-pub mod classes;
 pub mod concrete_type;
 pub mod expr;
 pub mod extern_fns;
@@ -43,12 +49,27 @@ pub mod irrt;
 pub mod llvm_intrinsics;
 pub mod numpy;
 pub mod stmt;
+pub mod types;
+pub mod values;
 
 #[cfg(test)]
 mod test;
 
-use concrete_type::{ConcreteType, ConcreteTypeEnum, ConcreteTypeStore};
-pub use generator::{CodeGenerator, DefaultCodeGenerator};
+mod macros {
+    /// Codegen-variant of [`std::unreachable`] which accepts an instance of [`CodeGenContext`] as
+    /// its first argument to provide Python source information to indicate the codegen location
+    /// causing the assertion.
+    macro_rules! codegen_unreachable {
+        ($ctx:expr $(,)?) => {
+            std::unreachable!("unreachable code while processing {}", &$ctx.current_loc)
+        };
+        ($ctx:expr, $($arg:tt)*) => {
+            std::unreachable!("unreachable code while processing {}: {}", &$ctx.current_loc, std::format!("{}", std::format_args!($($arg)+)))
+        };
+    }
+
+    pub(crate) use codegen_unreachable;
+}
 
 #[derive(Default)]
 pub struct StaticValueStore {
@@ -68,6 +89,16 @@ pub struct CodeGenLLVMOptions {
     pub target: CodeGenTargetMachineOptions,
 }
 
+impl CodeGenLLVMOptions {
+    /// Creates a [`TargetMachine`] using the target options specified by this struct.
+    ///
+    /// See [`Target::create_target_machine`].
+    #[must_use]
+    pub fn create_target_machine(&self) -> Option<TargetMachine> {
+        self.target.create_target_machine(self.opt_level)
+    }
+}
+
 /// Additional options for code generation for the target machine.
 #[derive(Clone, Debug, Eq, PartialEq)]
 pub struct CodeGenTargetMachineOptions {
@@ -338,6 +369,10 @@ impl WorkerRegistry {
         let mut builder = context.create_builder();
         let mut module = context.create_module(generator.get_name());
 
+        let target_machine = self.llvm_options.create_target_machine().unwrap();
+        module.set_data_layout(&target_machine.get_target_data().get_data_layout());
+        module.set_triple(&target_machine.get_triple());
+
         module.add_basic_value_flag(
             "Debug Info Version",
             inkwell::module::FlagBehavior::Warning,
@@ -361,6 +396,10 @@ impl WorkerRegistry {
                     errors.insert(e);
                     // create a new empty module just to continue codegen and collect errors
                     module = context.create_module(&format!("{}_recover", generator.get_name()));
+
+                    let target_machine = self.llvm_options.create_target_machine().unwrap();
+                    module.set_data_layout(&target_machine.get_target_data().get_data_layout());
+                    module.set_triple(&target_machine.get_triple());
                 }
             }
             *self.task_count.lock() -= 1;
@@ -426,7 +465,7 @@ pub struct CodeGenTask {
 fn get_llvm_type<'ctx, G: CodeGenerator + ?Sized>(
     ctx: &'ctx Context,
     module: &Module<'ctx>,
-    generator: &mut G,
+    generator: &G,
     unifier: &mut Unifier,
     top_level: &TopLevelContext,
     type_cache: &mut HashMap<Type, BasicTypeEnum<'ctx>>,
@@ -520,8 +559,10 @@ fn get_llvm_type<'ctx, G: CodeGenerator + ?Sized>(
                 };
                 return ty;
             }
-            TTuple { ty } => {
+            TTuple { ty, is_vararg_ctx } => {
                 // a struct with fields in the order present in the tuple
+                assert!(!is_vararg_ctx, "Tuples in vararg context must be instantiated with the correct number of arguments before calling get_llvm_type");
+
                 let fields = ty
                     .iter()
                     .map(|ty| {
@@ -551,7 +592,7 @@ fn get_llvm_type<'ctx, G: CodeGenerator + ?Sized>(
 fn get_llvm_abi_type<'ctx, G: CodeGenerator + ?Sized>(
     ctx: &'ctx Context,
     module: &Module<'ctx>,
-    generator: &mut G,
+    generator: &G,
     unifier: &mut Unifier,
     top_level: &TopLevelContext,
     type_cache: &mut HashMap<Type, BasicTypeEnum<'ctx>>,
@@ -560,11 +601,11 @@ fn get_llvm_abi_type<'ctx, G: CodeGenerator + ?Sized>(
 ) -> BasicTypeEnum<'ctx> {
     // If the type is used in the definition of a function, return `i1` instead of `i8` for ABI
     // consistency.
-    return if unifier.unioned(ty, primitives.bool) {
+    if unifier.unioned(ty, primitives.bool) {
         ctx.bool_type().into()
     } else {
         get_llvm_type(ctx, module, generator, unifier, top_level, type_cache, ty)
-    };
+    }
 }
 
 /// Whether `sret` is needed for a return value with type `ty`.
@@ -589,6 +630,40 @@ fn need_sret(ty: BasicTypeEnum) -> bool {
     need_sret_impl(ty, true)
 }
 
+/// Returns the [`BasicTypeEnum`] representing a `va_list` struct for variadic arguments.
+fn get_llvm_valist_type<'ctx>(ctx: &'ctx Context, triple: &TargetTriple) -> BasicTypeEnum<'ctx> {
+    let triple = TargetMachine::normalize_triple(triple);
+    let triple = triple.as_str().to_str().unwrap();
+    let arch = triple.split('-').next().unwrap();
+
+    let llvm_pi8 = ctx.i8_type().ptr_type(AddressSpace::default());
+
+    // Referenced from parseArch() in llvm/lib/Support/Triple.cpp
+    match arch {
+        "i386" | "i486" | "i586" | "i686" | "riscv32" => {
+            ctx.i8_type().ptr_type(AddressSpace::default()).into()
+        }
+        "amd64" | "x86_64" | "x86_64h" => {
+            let llvm_i32 = ctx.i32_type();
+
+            let va_list_tag = ctx.opaque_struct_type("struct.__va_list_tag");
+            va_list_tag.set_body(
+                &[llvm_i32.into(), llvm_i32.into(), llvm_pi8.into(), llvm_pi8.into()],
+                false,
+            );
+            va_list_tag.into()
+        }
+        "armv7" => {
+            let va_list = ctx.opaque_struct_type("struct.__va_list");
+            va_list.set_body(&[llvm_pi8.into()], false);
+            va_list.into()
+        }
+        triple => {
+            todo!("Unsupported platform for varargs: {triple}")
+        }
+    }
+}
+
 /// Implementation for generating LLVM IR for a function.
 pub fn gen_func_impl<
     'ctx,
@@ -700,6 +775,7 @@ pub fn gen_func_impl<
                 name: arg.name,
                 ty: task.store.to_unifier_type(&mut unifier, &primitives, arg.ty, &mut cache),
                 default_value: arg.default_value.clone(),
+                is_vararg: arg.is_vararg,
             })
             .collect_vec(),
         task.store.to_unifier_type(&mut unifier, &primitives, *ret, &mut cache),
@@ -722,7 +798,10 @@ pub fn gen_func_impl<
     let has_sret = ret_type.map_or(false, |ty| need_sret(ty));
     let mut params = args
         .iter()
+        .filter(|arg| !arg.is_vararg)
         .map(|arg| {
+            debug_assert!(!arg.is_vararg);
+
             get_llvm_abi_type(
                 context,
                 &module,
@@ -741,9 +820,12 @@ pub fn gen_func_impl<
         params.insert(0, ret_type.unwrap().ptr_type(AddressSpace::default()).into());
     }
 
+    debug_assert!(matches!(args.iter().filter(|arg| arg.is_vararg).count(), 0..=1));
+    let vararg_arg = args.iter().find(|arg| arg.is_vararg);
+
     let fn_type = match ret_type {
-        Some(ret_type) if !has_sret => ret_type.fn_type(&params, false),
-        _ => context.void_type().fn_type(&params, false),
+        Some(ret_type) if !has_sret => ret_type.fn_type(&params, vararg_arg.is_some()),
+        _ => context.void_type().fn_type(&params, vararg_arg.is_some()),
     };
 
     let symbol = &task.symbol_name;
@@ -771,9 +853,10 @@ pub fn gen_func_impl<
     builder.position_at_end(init_bb);
     let body_bb = context.append_basic_block(fn_val, "body");
 
+    // Store non-vararg argument values into local variables
     let mut var_assignment = HashMap::new();
     let offset = u32::from(has_sret);
-    for (n, arg) in args.iter().enumerate() {
+    for (n, arg) in args.iter().enumerate().filter(|(_, arg)| !arg.is_vararg) {
         let param = fn_val.get_nth_param((n as u32) + offset).unwrap();
         let local_type = get_llvm_type(
             context,
@@ -806,6 +889,8 @@ pub fn gen_func_impl<
         var_assignment.insert(arg.name, (alloca, None, 0));
     }
 
+    // TODO: Save vararg parameters as list
+
     let return_buffer = if has_sret {
         Some(fn_val.get_nth_param(0).unwrap().into_pointer_value())
     } else {
@@ -1028,3 +1113,9 @@ fn gen_in_range_check<'ctx>(
 
     ctx.builder.build_int_compare(IntPredicate::SLT, lo, hi, "cmp").unwrap()
 }
+
+/// Returns the internal name for the `va_count` argument, used to indicate the number of arguments
+/// passed to the variadic function.
+fn get_va_count_arg_name(arg_name: StrRef) -> StrRef {
+    format!("__{}_va_count", &arg_name).into()
+}

nac3core/src/codegen/stmt.rs

@@ -1,21 +1,3 @@
-use super::{
-    super::symbol_resolver::ValueEnum,
-    expr::destructure_range,
-    irrt::{handle_slice_indices, list_slice_assignment},
-    CodeGenContext, CodeGenerator,
-};
-use crate::{
-    codegen::{
-        classes::{ArrayLikeIndexer, ArraySliceValue, ListValue, RangeValue},
-        expr::gen_binop_expr,
-        gen_in_range_check,
-    },
-    toplevel::{helper::PrimDef, numpy::unpack_ndarray_var_tys, DefinitionId, TopLevelDef},
-    typecheck::{
-        magic_methods::Binop,
-        typedef::{FunSignature, Type, TypeEnum},
-    },
-};
 use inkwell::{
     attributes::{Attribute, AttributeLoc},
     basic_block::BasicBlock,
@@ -23,10 +5,28 @@ use inkwell::{
     values::{BasicValue, BasicValueEnum, FunctionValue, IntValue, PointerValue},
     IntPredicate,
 };
+use itertools::{izip, Itertools};
+
 use nac3parser::ast::{
     Constant, ExcepthandlerKind, Expr, ExprKind, Location, Stmt, StmtKind, StrRef,
 };
-use std::convert::TryFrom;
+
+use super::{
+    expr::{destructure_range, gen_binop_expr},
+    gen_in_range_check,
+    irrt::{handle_slice_indices, list_slice_assignment},
+    macros::codegen_unreachable,
+    values::{ArrayLikeIndexer, ArraySliceValue, ListValue, RangeValue},
+    CodeGenContext, CodeGenerator,
+};
+use crate::{
+    symbol_resolver::ValueEnum,
+    toplevel::{DefinitionId, TopLevelDef},
+    typecheck::{
+        magic_methods::Binop,
+        typedef::{iter_type_vars, FunSignature, Type, TypeEnum},
+    },
+};
 
 /// See [`CodeGenerator::gen_var_alloc`].
 pub fn gen_var<'ctx>(
@@ -97,8 +97,6 @@ pub fn gen_store_target<'ctx, G: CodeGenerator>(
     pattern: &Expr<Option<Type>>,
     name: Option<&str>,
 ) -> Result<Option<PointerValue<'ctx>>, String> {
-    let llvm_usize = generator.get_size_type(ctx.ctx);
-
     // very similar to gen_expr, but we don't do an extra load at the end
     // and we flatten nested tuples
     Ok(Some(match &pattern.node {
@@ -123,7 +121,7 @@ pub fn gen_store_target<'ctx, G: CodeGenerator>(
                 return Ok(None);
             };
             let BasicValueEnum::PointerValue(ptr) = val else {
-                unreachable!();
+                codegen_unreachable!(ctx);
             };
             unsafe {
                 ctx.builder.build_in_bounds_gep(
@@ -137,66 +135,7 @@ pub fn gen_store_target<'ctx, G: CodeGenerator>(
             }
             .unwrap()
         }
-        ExprKind::Subscript { value, slice, .. } => {
-            match ctx.unifier.get_ty_immutable(value.custom.unwrap()).as_ref() {
-                TypeEnum::TObj { obj_id, .. } if *obj_id == PrimDef::List.id() => {
-                    let v = generator
-                        .gen_expr(ctx, value)?
-                        .unwrap()
-                        .to_basic_value_enum(ctx, generator, value.custom.unwrap())?
-                        .into_pointer_value();
-                    let v = ListValue::from_ptr_val(v, llvm_usize, None);
-                    let len = v.load_size(ctx, Some("len"));
-                    let raw_index = generator
-                        .gen_expr(ctx, slice)?
-                        .unwrap()
-                        .to_basic_value_enum(ctx, generator, slice.custom.unwrap())?
-                        .into_int_value();
-                    let raw_index = ctx
-                        .builder
-                        .build_int_s_extend(raw_index, generator.get_size_type(ctx.ctx), "sext")
-                        .unwrap();
-                    // handle negative index
-                    let is_negative = ctx
-                        .builder
-                        .build_int_compare(
-                            IntPredicate::SLT,
-                            raw_index,
-                            generator.get_size_type(ctx.ctx).const_zero(),
-                            "is_neg",
-                        )
-                        .unwrap();
-                    let adjusted = ctx.builder.build_int_add(raw_index, len, "adjusted").unwrap();
-                    let index = ctx
-                        .builder
-                        .build_select(is_negative, adjusted, raw_index, "index")
-                        .map(BasicValueEnum::into_int_value)
-                        .unwrap();
-                    // unsigned less than is enough, because negative index after adjustment is
-                    // bigger than the length (for unsigned cmp)
-                    let bound_check = ctx
-                        .builder
-                        .build_int_compare(IntPredicate::ULT, index, len, "inbound")
-                        .unwrap();
-                    ctx.make_assert(
-                        generator,
-                        bound_check,
-                        "0:IndexError",
-                        "index {0} out of bounds 0:{1}",
-                        [Some(raw_index), Some(len), None],
-                        slice.location,
-                    );
-                    v.data().ptr_offset(ctx, generator, &index, name)
-                }
-
-                TypeEnum::TObj { obj_id, .. } if *obj_id == PrimDef::NDArray.id() => {
-                    todo!()
-                }
-
-                _ => unreachable!(),
-            }
-        }
-        _ => unreachable!(),
+        _ => codegen_unreachable!(ctx),
     }))
 }
 
@@ -206,70 +145,20 @@ pub fn gen_assign<'ctx, G: CodeGenerator>(
     ctx: &mut CodeGenContext<'ctx, '_>,
     target: &Expr<Option<Type>>,
     value: ValueEnum<'ctx>,
+    value_ty: Type,
 ) -> Result<(), String> {
-    let llvm_usize = generator.get_size_type(ctx.ctx);
-
+    // See https://docs.python.org/3/reference/simple_stmts.html#assignment-statements.
     match &target.node {
-        ExprKind::Tuple { elts, .. } => {
-            let BasicValueEnum::StructValue(v) =
-                value.to_basic_value_enum(ctx, generator, target.custom.unwrap())?
-            else {
-                unreachable!()
-            };
-
-            for (i, elt) in elts.iter().enumerate() {
-                let v = ctx
-                    .builder
-                    .build_extract_value(v, u32::try_from(i).unwrap(), "struct_elem")
-                    .unwrap();
-                generator.gen_assign(ctx, elt, v.into())?;
+        ExprKind::Subscript { value: target, slice: key, .. } => {
+            // Handle "slicing" or "subscription"
+            generator.gen_setitem(ctx, target, key, value, value_ty)?;
         }
-        }
-        ExprKind::Subscript { value: ls, slice, .. }
-            if matches!(&slice.node, ExprKind::Slice { .. }) =>
-        {
-            let ExprKind::Slice { lower, upper, step } = &slice.node else { unreachable!() };
-
-            let ls = generator
-                .gen_expr(ctx, ls)?
-                .unwrap()
-                .to_basic_value_enum(ctx, generator, ls.custom.unwrap())?
-                .into_pointer_value();
-            let ls = ListValue::from_ptr_val(ls, llvm_usize, None);
-            let Some((start, end, step)) =
-                handle_slice_indices(lower, upper, step, ctx, generator, ls.load_size(ctx, None))?
-            else {
-                return Ok(());
-            };
-            let value = value
-                .to_basic_value_enum(ctx, generator, target.custom.unwrap())?
-                .into_pointer_value();
-            let value = ListValue::from_ptr_val(value, llvm_usize, None);
-            let ty = match &*ctx.unifier.get_ty_immutable(target.custom.unwrap()) {
-                TypeEnum::TObj { obj_id, params, .. } if *obj_id == PrimDef::List.id() => {
-                    *params.iter().next().unwrap().1
-                }
-                TypeEnum::TObj { obj_id, .. } if *obj_id == PrimDef::NDArray.id() => {
-                    unpack_ndarray_var_tys(&mut ctx.unifier, target.custom.unwrap()).0
-                }
-                _ => unreachable!(),
-            };
-
-            let ty = ctx.get_llvm_type(generator, ty);
-            let Some(src_ind) = handle_slice_indices(
-                &None,
-                &None,
-                &None,
-                ctx,
-                generator,
-                value.load_size(ctx, None),
-            )?
-            else {
-                return Ok(());
-            };
-            list_slice_assignment(generator, ctx, ty, ls, (start, end, step), value, src_ind);
+        ExprKind::Tuple { elts, .. } | ExprKind::List { elts, .. } => {
+            // Fold on `"[" [target_list] "]"` and `"(" [target_list] ")"`
+            generator.gen_assign_target_list(ctx, elts, value, value_ty)?;
         }
         _ => {
+            // Handle attribute and direct variable assignments.
             let name = if let ExprKind::Name { id, .. } = &target.node {
                 format!("{id}.addr")
             } else {
@@ -287,19 +176,259 @@ pub fn gen_assign<'ctx, G: CodeGenerator>(
                 }
             }
             let val = value.to_basic_value_enum(ctx, generator, target.custom.unwrap())?;
+
+            // Perform i1 <-> i8 conversion as needed
+            let val = if ctx.unifier.unioned(target.custom.unwrap(), ctx.primitives.bool) {
+                generator.bool_to_i8(ctx, val.into_int_value()).into()
+            } else {
+                val
+            };
+
             ctx.builder.build_store(ptr, val).unwrap();
         }
     };
     Ok(())
 }
 
+/// See [`CodeGenerator::gen_assign_target_list`].
+pub fn gen_assign_target_list<'ctx, G: CodeGenerator>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    targets: &Vec<Expr<Option<Type>>>,
+    value: ValueEnum<'ctx>,
+    value_ty: Type,
+) -> Result<(), String> {
+    // Deconstruct the tuple `value`
+    let BasicValueEnum::StructValue(tuple) = value.to_basic_value_enum(ctx, generator, value_ty)?
+    else {
+        codegen_unreachable!(ctx)
+    };
+
+    // NOTE: Currently, RHS's type is forced to be a Tuple by the type inferencer.
+    let TypeEnum::TTuple { ty: tuple_tys, .. } = &*ctx.unifier.get_ty(value_ty) else {
+        codegen_unreachable!(ctx);
+    };
+
+    assert_eq!(tuple.get_type().count_fields() as usize, tuple_tys.len());
+
+    let tuple = (0..tuple.get_type().count_fields())
+        .map(|i| ctx.builder.build_extract_value(tuple, i, "item").unwrap())
+        .collect_vec();
+
+    // Find the starred target if it exists.
+    let mut starred_target_index: Option<usize> = None; // Index of the "starred" target. If it exists, there may only be one.
+    for (i, target) in targets.iter().enumerate() {
+        if matches!(target.node, ExprKind::Starred { .. }) {
+            assert!(starred_target_index.is_none()); // The typechecker ensures this
+            starred_target_index = Some(i);
+        }
+    }
+
+    if let Some(starred_target_index) = starred_target_index {
+        assert!(tuple_tys.len() >= targets.len() - 1); // The typechecker ensures this
+
+        let a = starred_target_index; // Number of RHS values before the starred target
+        let b = tuple_tys.len() - (targets.len() - 1 - starred_target_index); // Number of RHS values after the starred target
+                                                                              // Thus `tuple[a..b]` is assigned to the starred target.
+
+        // Handle assignment before the starred target
+        for (target, val, val_ty) in
+            izip!(&targets[..starred_target_index], &tuple[..a], &tuple_tys[..a])
+        {
+            generator.gen_assign(ctx, target, ValueEnum::Dynamic(*val), *val_ty)?;
+        }
+
+        // Handle assignment to the starred target
+        if let ExprKind::Starred { value: target, .. } = &targets[starred_target_index].node {
+            let vals = &tuple[a..b];
+            let val_tys = &tuple_tys[a..b];
+
+            // Create a sub-tuple from `value` for the starred target.
+            let sub_tuple_ty = ctx
+                .ctx
+                .struct_type(&vals.iter().map(BasicValueEnum::get_type).collect_vec(), false);
+            let psub_tuple_val =
+                ctx.builder.build_alloca(sub_tuple_ty, "starred_target_value_ptr").unwrap();
+            for (i, val) in vals.iter().enumerate() {
+                let pitem = ctx
+                    .builder
+                    .build_struct_gep(psub_tuple_val, i as u32, "starred_target_value_item")
+                    .unwrap();
+                ctx.builder.build_store(pitem, *val).unwrap();
+            }
+            let sub_tuple_val =
+                ctx.builder.build_load(psub_tuple_val, "starred_target_value").unwrap();
+
+            // Create the typechecker type of the sub-tuple
+            let sub_tuple_ty =
+                ctx.unifier.add_ty(TypeEnum::TTuple { ty: val_tys.to_vec(), is_vararg_ctx: false });
+
+            // Now assign with that sub-tuple to the starred target.
+            generator.gen_assign(ctx, target, ValueEnum::Dynamic(sub_tuple_val), sub_tuple_ty)?;
+        } else {
+            codegen_unreachable!(ctx) // The typechecker ensures this
+        }
+
+        // Handle assignment after the starred target
+        for (target, val, val_ty) in
+            izip!(&targets[starred_target_index + 1..], &tuple[b..], &tuple_tys[b..])
+        {
+            generator.gen_assign(ctx, target, ValueEnum::Dynamic(*val), *val_ty)?;
+        }
+    } else {
+        assert_eq!(tuple_tys.len(), targets.len()); // The typechecker ensures this
+
+        for (target, val, val_ty) in izip!(targets, tuple, tuple_tys) {
+            generator.gen_assign(ctx, target, ValueEnum::Dynamic(val), *val_ty)?;
+        }
+    }
+    Ok(())
+}
+
+/// See [`CodeGenerator::gen_setitem`].
+pub fn gen_setitem<'ctx, G: CodeGenerator>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    target: &Expr<Option<Type>>,
+    key: &Expr<Option<Type>>,
+    value: ValueEnum<'ctx>,
+    value_ty: Type,
+) -> Result<(), String> {
+    let target_ty = target.custom.unwrap();
+    let key_ty = key.custom.unwrap();
+
+    match &*ctx.unifier.get_ty(target_ty) {
+        TypeEnum::TObj { obj_id, params: list_params, .. }
+            if *obj_id == ctx.primitives.list.obj_id(&ctx.unifier).unwrap() =>
+        {
+            // Handle list item assignment
+            let llvm_usize = generator.get_size_type(ctx.ctx);
+            let target_item_ty = iter_type_vars(list_params).next().unwrap().ty;
+
+            let target = generator
+                .gen_expr(ctx, target)?
+                .unwrap()
+                .to_basic_value_enum(ctx, generator, target_ty)?
+                .into_pointer_value();
+            let target = ListValue::from_pointer_value(target, llvm_usize, None);
+
+            if let ExprKind::Slice { .. } = &key.node {
+                // Handle assigning to a slice
+                let ExprKind::Slice { lower, upper, step } = &key.node else {
+                    codegen_unreachable!(ctx)
+                };
+                let Some((start, end, step)) = handle_slice_indices(
+                    lower,
+                    upper,
+                    step,
+                    ctx,
+                    generator,
+                    target.load_size(ctx, None),
+                )?
+                else {
+                    return Ok(());
+                };
+
+                let value =
+                    value.to_basic_value_enum(ctx, generator, value_ty)?.into_pointer_value();
+                let value = ListValue::from_pointer_value(value, llvm_usize, None);
+
+                let target_item_ty = ctx.get_llvm_type(generator, target_item_ty);
+                let Some(src_ind) = handle_slice_indices(
+                    &None,
+                    &None,
+                    &None,
+                    ctx,
+                    generator,
+                    value.load_size(ctx, None),
+                )?
+                else {
+                    return Ok(());
+                };
+                list_slice_assignment(
+                    generator,
+                    ctx,
+                    target_item_ty,
+                    target,
+                    (start, end, step),
+                    value,
+                    src_ind,
+                );
+            } else {
+                // Handle assigning to an index
+                let len = target.load_size(ctx, Some("len"));
+
+                let index = generator
+                    .gen_expr(ctx, key)?
+                    .unwrap()
+                    .to_basic_value_enum(ctx, generator, key_ty)?
+                    .into_int_value();
+                let index = ctx
+                    .builder
+                    .build_int_s_extend(index, generator.get_size_type(ctx.ctx), "sext")
+                    .unwrap();
+
+                // handle negative index
+                let is_negative = ctx
+                    .builder
+                    .build_int_compare(
+                        IntPredicate::SLT,
+                        index,
+                        generator.get_size_type(ctx.ctx).const_zero(),
+                        "is_neg",
+                    )
+                    .unwrap();
+                let adjusted = ctx.builder.build_int_add(index, len, "adjusted").unwrap();
+                let index = ctx
+                    .builder
+                    .build_select(is_negative, adjusted, index, "index")
+                    .map(BasicValueEnum::into_int_value)
+                    .unwrap();
+
+                // unsigned less than is enough, because negative index after adjustment is
+                // bigger than the length (for unsigned cmp)
+                let bound_check = ctx
+                    .builder
+                    .build_int_compare(IntPredicate::ULT, index, len, "inbound")
+                    .unwrap();
+                ctx.make_assert(
+                    generator,
+                    bound_check,
+                    "0:IndexError",
+                    "index {0} out of bounds 0:{1}",
+                    [Some(index), Some(len), None],
+                    key.location,
+                );
+
+                // Write value to index on list
+                let item_ptr =
+                    target.data().ptr_offset(ctx, generator, &index, Some("list_item_ptr"));
+                let value = value.to_basic_value_enum(ctx, generator, value_ty)?;
+                ctx.builder.build_store(item_ptr, value).unwrap();
+            }
+        }
+        TypeEnum::TObj { obj_id, .. }
+            if *obj_id == ctx.primitives.ndarray.obj_id(&ctx.unifier).unwrap() =>
+        {
+            // Handle NDArray item assignment
+            todo!("ndarray subscript assignment is not yet implemented");
+        }
+        _ => {
+            panic!("encountered unknown target type: {}", ctx.unifier.stringify(target_ty));
+        }
+    }
+    Ok(())
+}
+
 /// See [`CodeGenerator::gen_for`].
 pub fn gen_for<G: CodeGenerator>(
     generator: &mut G,
     ctx: &mut CodeGenContext<'_, '_>,
     stmt: &Stmt<Option<Type>>,
 ) -> Result<(), String> {
-    let StmtKind::For { iter, target, body, orelse, .. } = &stmt.node else { unreachable!() };
+    let StmtKind::For { iter, target, body, orelse, .. } = &stmt.node else {
+        codegen_unreachable!(ctx)
+    };
 
     // var_assignment static values may be changed in another branch
     // if so, remove the static value as it may not be correct in this branch
@@ -315,9 +444,6 @@ pub fn gen_for<G: CodeGenerator>(
     let orelse_bb =
         if orelse.is_empty() { cont_bb } else { ctx.ctx.append_basic_block(current, "for.orelse") };
 
-    // Whether the iterable is a range() expression
-    let is_iterable_range_expr = ctx.unifier.unioned(iter.custom.unwrap(), ctx.primitives.range);
-
     // The BB containing the increment expression
     let incr_bb = ctx.ctx.append_basic_block(current, "for.incr");
     // The BB containing the loop condition check
@@ -326,27 +452,36 @@ pub fn gen_for<G: CodeGenerator>(
     // store loop bb information and restore it later
     let loop_bb = ctx.loop_target.replace((incr_bb, cont_bb));
 
+    let iter_ty = iter.custom.unwrap();
     let iter_val = if let Some(v) = generator.gen_expr(ctx, iter)? {
-        v.to_basic_value_enum(ctx, generator, iter.custom.unwrap())?
+        v.to_basic_value_enum(ctx, generator, iter_ty)?
     } else {
         return Ok(());
     };
-    if is_iterable_range_expr {
-        let iter_val = RangeValue::from_ptr_val(iter_val.into_pointer_value(), Some("range"));
+
+    match &*ctx.unifier.get_ty(iter_ty) {
+        TypeEnum::TObj { obj_id, .. }
+            if *obj_id == ctx.primitives.range.obj_id(&ctx.unifier).unwrap() =>
+        {
+            let iter_val =
+                RangeValue::from_pointer_value(iter_val.into_pointer_value(), Some("range"));
             // Internal variable for loop; Cannot be assigned
             let i = generator.gen_var_alloc(ctx, int32.into(), Some("for.i.addr"))?;
             // Variable declared in "target" expression of the loop; Can be reassigned *or* shadowed
-        let Some(target_i) = generator.gen_store_target(ctx, target, Some("for.target.addr"))?
+            let Some(target_i) =
+                generator.gen_store_target(ctx, target, Some("for.target.addr"))?
             else {
-            unreachable!()
+                codegen_unreachable!(ctx)
             };
             let (start, stop, step) = destructure_range(ctx, iter_val);
 
             ctx.builder.build_store(i, start).unwrap();
 
             // Check "If step is zero, ValueError is raised."
-        let rangenez =
-            ctx.builder.build_int_compare(IntPredicate::NE, step, int32.const_zero(), "").unwrap();
+            let rangenez = ctx
+                .builder
+                .build_int_compare(IntPredicate::NE, step, int32.const_zero(), "")
+                .unwrap();
             ctx.make_assert(
                 generator,
                 rangenez,
@@ -363,7 +498,10 @@ pub fn gen_for<G: CodeGenerator>(
                     .build_conditional_branch(
                         gen_in_range_check(
                             ctx,
-                        ctx.builder.build_load(i, "").map(BasicValueEnum::into_int_value).unwrap(),
+                            ctx.builder
+                                .build_load(i, "")
+                                .map(BasicValueEnum::into_int_value)
+                                .unwrap(),
                             stop,
                             step,
                         ),
@@ -393,7 +531,10 @@ pub fn gen_for<G: CodeGenerator>(
                 )
                 .unwrap();
             generator.gen_block(ctx, body.iter())?;
-    } else {
+        }
+        TypeEnum::TObj { obj_id, params: list_params, .. }
+            if *obj_id == ctx.primitives.list.obj_id(&ctx.unifier).unwrap() =>
+        {
             let index_addr = generator.gen_var_alloc(ctx, size_t.into(), Some("for.index.addr"))?;
             ctx.builder.build_store(index_addr, size_t.const_zero()).unwrap();
             let len = ctx
@@ -431,9 +572,14 @@ pub fn gen_for<G: CodeGenerator>(
                 .map(BasicValueEnum::into_int_value)
                 .unwrap();
             let val = ctx.build_gep_and_load(arr_ptr, &[index], Some("val"));
-        generator.gen_assign(ctx, target, val.into())?;
+            let val_ty = iter_type_vars(list_params).next().unwrap().ty;
+            generator.gen_assign(ctx, target, val.into(), val_ty)?;
             generator.gen_block(ctx, body.iter())?;
         }
+        _ => {
+            panic!("unsupported for loop iterator type: {}", ctx.unifier.stringify(iter_ty));
+        }
+    }
 
     for (k, (_, _, counter)) in &var_assignment {
         let (_, static_val, counter2) = ctx.var_assignment.get_mut(k).unwrap();
@@ -494,6 +640,7 @@ pub struct BreakContinueHooks<'ctx> {
 pub fn gen_for_callback<'ctx, 'a, G, I, InitFn, CondFn, BodyFn, UpdateFn>(
     generator: &mut G,
     ctx: &mut CodeGenContext<'ctx, 'a>,
+    label: Option<&str>,
     init: InitFn,
     cond: CondFn,
     body: BodyFn,
@@ -504,18 +651,24 @@ where
     I: Clone,
     InitFn: FnOnce(&mut G, &mut CodeGenContext<'ctx, 'a>) -> Result<I, String>,
     CondFn: FnOnce(&mut G, &mut CodeGenContext<'ctx, 'a>, I) -> Result<IntValue<'ctx>, String>,
-    BodyFn:
-        FnOnce(&mut G, &mut CodeGenContext<'ctx, 'a>, BreakContinueHooks, I) -> Result<(), String>,
+    BodyFn: FnOnce(
+        &mut G,
+        &mut CodeGenContext<'ctx, 'a>,
+        BreakContinueHooks<'ctx>,
+        I,
+    ) -> Result<(), String>,
     UpdateFn: FnOnce(&mut G, &mut CodeGenContext<'ctx, 'a>, I) -> Result<(), String>,
 {
+    let label = label.unwrap_or("for");
+
     let current_bb = ctx.builder.get_insert_block().unwrap();
-    let init_bb = ctx.ctx.insert_basic_block_after(current_bb, "for.init");
+    let init_bb = ctx.ctx.insert_basic_block_after(current_bb, &format!("{label}.init"));
     // The BB containing the loop condition check
-    let cond_bb = ctx.ctx.insert_basic_block_after(init_bb, "for.cond");
-    let body_bb = ctx.ctx.insert_basic_block_after(cond_bb, "for.body");
+    let cond_bb = ctx.ctx.insert_basic_block_after(init_bb, &format!("{label}.cond"));
+    let body_bb = ctx.ctx.insert_basic_block_after(cond_bb, &format!("{label}.body"));
     // The BB containing the increment expression
-    let update_bb = ctx.ctx.insert_basic_block_after(body_bb, "for.update");
-    let cont_bb = ctx.ctx.insert_basic_block_after(update_bb, "for.end");
+    let update_bb = ctx.ctx.insert_basic_block_after(body_bb, &format!("{label}.update"));
+    let cont_bb = ctx.ctx.insert_basic_block_after(update_bb, &format!("{label}.end"));
 
     // store loop bb information and restore it later
     let loop_bb = ctx.loop_target.replace((update_bb, cont_bb));
@@ -572,6 +725,7 @@ where
 pub fn gen_for_callback_incrementing<'ctx, 'a, G, BodyFn>(
     generator: &mut G,
     ctx: &mut CodeGenContext<'ctx, 'a>,
+    label: Option<&str>,
     init_val: IntValue<'ctx>,
     max_val: (IntValue<'ctx>, bool),
     body: BodyFn,
@@ -582,7 +736,7 @@ where
     BodyFn: FnOnce(
         &mut G,
         &mut CodeGenContext<'ctx, 'a>,
-        BreakContinueHooks,
+        BreakContinueHooks<'ctx>,
         IntValue<'ctx>,
     ) -> Result<(), String>,
 {
@@ -591,6 +745,7 @@ where
     gen_for_callback(
         generator,
         ctx,
+        label,
         |generator, ctx| {
             let i_addr = generator.gen_var_alloc(ctx, init_val_t.into(), None)?;
             ctx.builder.build_store(i_addr, init_val).unwrap();
@@ -642,9 +797,11 @@ where
 /// - `step_fn`: A lambda of IR statements that retrieves the `step` value of the  `range`-like
 ///   iterable. This value will be extended to the size of `start`.
 /// - `body_fn`: A lambda of IR statements within the loop body.
+#[allow(clippy::too_many_arguments)]
 pub fn gen_for_range_callback<'ctx, 'a, G, StartFn, StopFn, StepFn, BodyFn>(
     generator: &mut G,
     ctx: &mut CodeGenContext<'ctx, 'a>,
+    label: Option<&str>,
     is_unsigned: bool,
     start_fn: StartFn,
     (stop_fn, stop_inclusive): (StopFn, bool),
@@ -656,13 +813,19 @@ where
     StartFn: Fn(&mut G, &mut CodeGenContext<'ctx, 'a>) -> Result<IntValue<'ctx>, String>,
     StopFn: Fn(&mut G, &mut CodeGenContext<'ctx, 'a>) -> Result<IntValue<'ctx>, String>,
     StepFn: Fn(&mut G, &mut CodeGenContext<'ctx, 'a>) -> Result<IntValue<'ctx>, String>,
-    BodyFn: FnOnce(&mut G, &mut CodeGenContext<'ctx, 'a>, IntValue<'ctx>) -> Result<(), String>,
+    BodyFn: FnOnce(
+        &mut G,
+        &mut CodeGenContext<'ctx, 'a>,
+        BreakContinueHooks<'ctx>,
+        IntValue<'ctx>,
+    ) -> Result<(), String>,
 {
     let init_val_t = start_fn(generator, ctx).map(IntValue::get_type).unwrap();
 
     gen_for_callback(
         generator,
         ctx,
+        label,
         |generator, ctx| {
             let i_addr = generator.gen_var_alloc(ctx, init_val_t.into(), None)?;
 
@@ -720,10 +883,10 @@ where
 
             Ok(cond)
         },
-        |generator, ctx, _, (i_addr, _)| {
+        |generator, ctx, hooks, (i_addr, _)| {
             let i = ctx.builder.build_load(i_addr, "").map(BasicValueEnum::into_int_value).unwrap();
 
-            body_fn(generator, ctx, i)
+            body_fn(generator, ctx, hooks, i)
         },
         |generator, ctx, (i_addr, _)| {
             let i = ctx.builder.build_load(i_addr, "").map(BasicValueEnum::into_int_value).unwrap();
@@ -751,7 +914,7 @@ pub fn gen_while<G: CodeGenerator>(
     ctx: &mut CodeGenContext<'_, '_>,
     stmt: &Stmt<Option<Type>>,
 ) -> Result<(), String> {
-    let StmtKind::While { test, body, orelse, .. } = &stmt.node else { unreachable!() };
+    let StmtKind::While { test, body, orelse, .. } = &stmt.node else { codegen_unreachable!(ctx) };
 
     // var_assignment static values may be changed in another branch
     // if so, remove the static value as it may not be correct in this branch
@@ -781,7 +944,7 @@ pub fn gen_while<G: CodeGenerator>(
 
         return Ok(());
     };
-    let BasicValueEnum::IntValue(test) = test else { unreachable!() };
+    let BasicValueEnum::IntValue(test) = test else { codegen_unreachable!(ctx) };
 
     ctx.builder
         .build_conditional_branch(generator.bool_to_i1(ctx, test), body_bb, orelse_bb)
@@ -929,7 +1092,7 @@ pub fn gen_if<G: CodeGenerator>(
     ctx: &mut CodeGenContext<'_, '_>,
     stmt: &Stmt<Option<Type>>,
 ) -> Result<(), String> {
-    let StmtKind::If { test, body, orelse, .. } = &stmt.node else { unreachable!() };
+    let StmtKind::If { test, body, orelse, .. } = &stmt.node else { codegen_unreachable!(ctx) };
 
     // var_assignment static values may be changed in another branch
     // if so, remove the static value as it may not be correct in this branch
@@ -1052,11 +1215,11 @@ pub fn exn_constructor<'ctx>(
     let zelf_id = if let TypeEnum::TObj { obj_id, .. } = &*ctx.unifier.get_ty(zelf_ty) {
         obj_id.0
     } else {
-        unreachable!()
+        codegen_unreachable!(ctx)
     };
     let defs = ctx.top_level.definitions.read();
     let def = defs[zelf_id].read();
-    let TopLevelDef::Class { name: zelf_name, .. } = &*def else { unreachable!() };
+    let TopLevelDef::Class { name: zelf_name, .. } = &*def else { codegen_unreachable!(ctx) };
     let exception_name = format!("{}:{}", ctx.resolver.get_exception_id(zelf_id), zelf_name);
     unsafe {
         let id_ptr = ctx.builder.build_in_bounds_gep(zelf, &[zero, zero], "exn.id").unwrap();
@@ -1164,7 +1327,7 @@ pub fn gen_try<'ctx, 'a, G: CodeGenerator>(
     target: &Stmt<Option<Type>>,
 ) -> Result<(), String> {
     let StmtKind::Try { body, handlers, orelse, finalbody, .. } = &target.node else {
-        unreachable!()
+        codegen_unreachable!(ctx)
     };
 
     // if we need to generate anything related to exception, we must have personality defined
@@ -1241,7 +1404,7 @@ pub fn gen_try<'ctx, 'a, G: CodeGenerator>(
             if let TypeEnum::TObj { obj_id, .. } = &*ctx.unifier.get_ty(type_.custom.unwrap()) {
                 *obj_id
             } else {
-                unreachable!()
+                codegen_unreachable!(ctx)
             };
         let exception_name = format!("{}:{}", ctx.resolver.get_exception_id(obj_id.0), exn_name);
         let exn_id = ctx.resolver.get_string_id(&exception_name);
@@ -1513,6 +1676,23 @@ pub fn gen_return<G: CodeGenerator>(
     } else {
         None
     };
+
+    // Remap boolean return type into i1
+    let value = value.map(|ret_val| {
+        // The "return type" of a sret function is in the first parameter
+        let expected_ty = if ctx.need_sret {
+            func.get_type().get_param_types()[0]
+        } else {
+            func.get_type().get_return_type().unwrap()
+        };
+
+        if matches!(expected_ty, BasicTypeEnum::IntType(ty) if ty.get_bit_width() == 1) {
+            generator.bool_to_i1(ctx, ret_val.into_int_value()).into()
+        } else {
+            ret_val
+        }
+    });
+
     if let Some(return_target) = ctx.return_target {
         if let Some(value) = value {
             ctx.builder.build_store(ctx.return_buffer.unwrap(), value).unwrap();
@@ -1523,25 +1703,6 @@ pub fn gen_return<G: CodeGenerator>(
         ctx.builder.build_store(ctx.return_buffer.unwrap(), value.unwrap()).unwrap();
         ctx.builder.build_return(None).unwrap();
     } else {
-        // Remap boolean return type into i1
-        let value = value.map(|v| {
-            let expected_ty = func.get_type().get_return_type().unwrap();
-            let ret_val = v.as_basic_value_enum();
-
-            if expected_ty.is_int_type() && ret_val.is_int_value() {
-                let ret_type = expected_ty.into_int_type();
-                let ret_val = ret_val.into_int_value();
-
-                if ret_type.get_bit_width() == 1 && ret_val.get_type().get_bit_width() != 1 {
-                    generator.bool_to_i1(ctx, ret_val)
-                } else {
-                    ret_val
-                }
-                .into()
-            } else {
-                ret_val
-            }
-        });
         let value = value.as_ref().map(|v| v as &dyn BasicValue);
         ctx.builder.build_return(value).unwrap();
     }
@@ -1575,14 +1736,14 @@ pub fn gen_stmt<G: CodeGenerator>(
         }
         StmtKind::AnnAssign { target, value, .. } => {
             if let Some(value) = value {
-                let Some(value) = generator.gen_expr(ctx, value)? else { return Ok(()) };
-                generator.gen_assign(ctx, target, value)?;
+                let Some(value_enum) = generator.gen_expr(ctx, value)? else { return Ok(()) };
+                generator.gen_assign(ctx, target, value_enum, value.custom.unwrap())?;
             }
         }
         StmtKind::Assign { targets, value, .. } => {
-            let Some(value) = generator.gen_expr(ctx, value)? else { return Ok(()) };
+            let Some(value_enum) = generator.gen_expr(ctx, value)? else { return Ok(()) };
             for target in targets {
-                generator.gen_assign(ctx, target, value.clone())?;
+                generator.gen_assign(ctx, target, value_enum.clone(), value.custom.unwrap())?;
             }
         }
         StmtKind::Continue { .. } => {
@@ -1596,20 +1757,44 @@ pub fn gen_stmt<G: CodeGenerator>(
         StmtKind::For { .. } => generator.gen_for(ctx, stmt)?,
         StmtKind::With { .. } => generator.gen_with(ctx, stmt)?,
         StmtKind::AugAssign { target, op, value, .. } => {
-            let value = gen_binop_expr(
+            let value_enum = gen_binop_expr(
                 generator,
                 ctx,
                 target,
                 Binop::aug_assign(*op),
                 value,
                 stmt.location,
-            )?;
-            generator.gen_assign(ctx, target, value.unwrap())?;
+            )?
+            .unwrap();
+            generator.gen_assign(ctx, target, value_enum, value.custom.unwrap())?;
         }
         StmtKind::Try { .. } => gen_try(generator, ctx, stmt)?,
         StmtKind::Raise { exc, .. } => {
             if let Some(exc) = exc {
-                let exc = if let Some(v) = generator.gen_expr(ctx, exc)? {
+                let exn = if let ExprKind::Name { id, .. } = &exc.node {
+                    // Handle "raise Exception" short form
+                    let def_id = ctx.resolver.get_identifier_def(*id).map_err(|e| {
+                        format!("{} (at {})", e.iter().next().unwrap(), exc.location)
+                    })?;
+                    let def = ctx.top_level.definitions.read();
+                    let TopLevelDef::Class { constructor, .. } = *def[def_id.0].read() else {
+                        return Err(format!("Failed to resolve symbol {id} (at {})", exc.location));
+                    };
+
+                    let TypeEnum::TFunc(signature) =
+                        ctx.unifier.get_ty(constructor.unwrap()).as_ref().clone()
+                    else {
+                        return Err(format!("Failed to resolve symbol {id} (at {})", exc.location));
+                    };
+
+                    generator
+                        .gen_call(ctx, None, (&signature, def_id), Vec::default())?
+                        .map(Into::into)
+                } else {
+                    generator.gen_expr(ctx, exc)?
+                };
+
+                let exc = if let Some(v) = exn {
                     v.to_basic_value_enum(ctx, generator, exc.custom.unwrap())?
                 } else {
                     return Ok(());
@@ -1633,17 +1818,48 @@ pub fn gen_stmt<G: CodeGenerator>(
                         return Ok(());
                     }
                 }
-                None => ctx.gen_string(generator, ""),
+                None => ctx.gen_string(generator, "").into(),
             };
             ctx.make_assert_impl(
                 generator,
-                test.into_int_value(),
+                generator.bool_to_i1(ctx, test.into_int_value()),
                 "0:AssertionError",
                 err_msg,
                 [None, None, None],
                 stmt.location,
             );
         }
+        StmtKind::Global { names, .. } => {
+            let registered_globals = ctx
+                .top_level
+                .definitions
+                .read()
+                .iter()
+                .filter_map(|def| {
+                    if let TopLevelDef::Variable { simple_name, ty, .. } = &*def.read() {
+                        Some((*simple_name, *ty))
+                    } else {
+                        None
+                    }
+                })
+                .collect_vec();
+
+            for id in names {
+                let Some((_, ty)) = registered_globals.iter().find(|(name, _)| name == id) else {
+                    return Err(format!("{id} is not a global at {}", stmt.location));
+                };
+
+                let resolver = ctx.resolver.clone();
+                let ptr = resolver
+                    .get_symbol_value(*id, ctx, generator)
+                    .map(|val| val.to_basic_value_enum(ctx, generator, *ty))
+                    .transpose()?
+                    .map(BasicValueEnum::into_pointer_value)
+                    .unwrap();
+
+                ctx.var_assignment.insert(*id, (ptr, None, 0));
+            }
+        }
         _ => unimplemented!(),
     };
     Ok(())

nac3core/src/codegen/test.rs

@@ -1,34 +1,37 @@
-use crate::{
-    codegen::{
-        classes::{ListType, NDArrayType, ProxyType, RangeType},
-        concrete_type::ConcreteTypeStore,
-        CodeGenContext, CodeGenLLVMOptions, CodeGenTargetMachineOptions, CodeGenTask,
-        CodeGenerator, DefaultCodeGenerator, WithCall, WorkerRegistry,
-    },
-    symbol_resolver::{SymbolResolver, ValueEnum},
-    toplevel::{
-        composer::{ComposerConfig, TopLevelComposer},
-        DefinitionId, FunInstance, TopLevelContext, TopLevelDef,
-    },
-    typecheck::{
-        type_inferencer::{FunctionData, Inferencer, PrimitiveStore},
-        typedef::{FunSignature, FuncArg, Type, TypeEnum, Unifier, VarMap},
-    },
+use std::{
+    collections::{HashMap, HashSet},
+    sync::Arc,
 };
+
 use indexmap::IndexMap;
 use indoc::indoc;
 use inkwell::{
     targets::{InitializationConfig, Target},
     OptimizationLevel,
 };
-use nac3parser::ast::FileName;
 use nac3parser::{
-    ast::{fold::Fold, StrRef},
+    ast::{fold::Fold, FileName, StrRef},
     parser::parse_program,
 };
 use parking_lot::RwLock;
-use std::collections::{HashMap, HashSet};
-use std::sync::Arc;
+
+use super::{
+    concrete_type::ConcreteTypeStore,
+    types::{ListType, NDArrayType, ProxyType, RangeType},
+    CodeGenContext, CodeGenLLVMOptions, CodeGenTargetMachineOptions, CodeGenTask, CodeGenerator,
+    DefaultCodeGenerator, WithCall, WorkerRegistry,
+};
+use crate::{
+    symbol_resolver::{SymbolResolver, ValueEnum},
+    toplevel::{
+        composer::{ComposerConfig, TopLevelComposer},
+        DefinitionId, FunInstance, TopLevelContext, TopLevelDef,
+    },
+    typecheck::{
+        type_inferencer::{FunctionData, IdentifierInfo, Inferencer, PrimitiveStore},
+        typedef::{FunSignature, FuncArg, Type, TypeEnum, Unifier, VarMap},
+    },
+};
 
 struct Resolver {
     id_to_type: HashMap<StrRef, Type>,
@@ -64,6 +67,7 @@ impl SymbolResolver for Resolver {
         &self,
         _: StrRef,
         _: &mut CodeGenContext<'ctx, '_>,
+        _: &mut dyn CodeGenerator,
     ) -> Option<ValueEnum<'ctx>> {
         unimplemented!()
     }
@@ -94,7 +98,7 @@ fn test_primitives() {
         "};
     let statements = parse_program(source, FileName::default()).unwrap();
 
-    let composer = TopLevelComposer::new(Vec::new(), ComposerConfig::default(), 32).0;
+    let composer = TopLevelComposer::new(Vec::new(), Vec::new(), ComposerConfig::default(), 32).0;
     let mut unifier = composer.unifier.clone();
     let primitives = composer.primitives_ty;
     let top_level = Arc::new(composer.make_top_level_context());
@@ -109,8 +113,18 @@ fn test_primitives() {
     let threads = vec![DefaultCodeGenerator::new("test".into(), 32).into()];
     let signature = FunSignature {
         args: vec![
-            FuncArg { name: "a".into(), ty: primitives.int32, default_value: None },
-            FuncArg { name: "b".into(), ty: primitives.int32, default_value: None },
+            FuncArg {
+                name: "a".into(),
+                ty: primitives.int32,
+                default_value: None,
+                is_vararg: false,
+            },
+            FuncArg {
+                name: "b".into(),
+                ty: primitives.int32,
+                default_value: None,
+                is_vararg: false,
+            },
         ],
         ret: primitives.int32,
         vars: VarMap::new(),
@@ -128,7 +142,8 @@ fn test_primitives() {
     };
     let mut virtual_checks = Vec::new();
     let mut calls = HashMap::new();
-    let mut identifiers: HashSet<_> = ["a".into(), "b".into()].into();
+    let mut identifiers: HashMap<_, _> =
+        ["a".into(), "b".into()].map(|id| (id, IdentifierInfo::default())).into();
     let mut inferencer = Inferencer {
         top_level: &top_level,
         function_data: &mut function_data,
@@ -189,6 +204,8 @@ fn test_primitives() {
         let expected = indoc! {"
             ; ModuleID = 'test'
             source_filename = \"test\"
+            target datalayout = \"e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128\"
+            target triple = \"x86_64-unknown-linux-gnu\"
 
             ; Function Attrs: mustprogress nofree norecurse nosync nounwind readnone willreturn
             define i32 @testing(i32 %0, i32 %1) local_unnamed_addr #0 !dbg !4 {
@@ -246,14 +263,19 @@ fn test_simple_call() {
         "};
     let statements_2 = parse_program(source_2, FileName::default()).unwrap();
 
-    let composer = TopLevelComposer::new(Vec::new(), ComposerConfig::default(), 32).0;
+    let composer = TopLevelComposer::new(Vec::new(), Vec::new(), ComposerConfig::default(), 32).0;
     let mut unifier = composer.unifier.clone();
     let primitives = composer.primitives_ty;
     let top_level = Arc::new(composer.make_top_level_context());
     unifier.top_level = Some(top_level.clone());
 
     let signature = FunSignature {
-        args: vec![FuncArg { name: "a".into(), ty: primitives.int32, default_value: None }],
+        args: vec![FuncArg {
+            name: "a".into(),
+            ty: primitives.int32,
+            default_value: None,
+            is_vararg: false,
+        }],
         ret: primitives.int32,
         vars: VarMap::new(),
     };
@@ -300,7 +322,8 @@ fn test_simple_call() {
     };
     let mut virtual_checks = Vec::new();
     let mut calls = HashMap::new();
-    let mut identifiers: HashSet<_> = ["a".into(), "foo".into()].into();
+    let mut identifiers: HashMap<_, _> =
+        ["a".into(), "foo".into()].map(|id| (id, IdentifierInfo::default())).into();
     let mut inferencer = Inferencer {
         top_level: &top_level,
         function_data: &mut function_data,
@@ -368,6 +391,8 @@ fn test_simple_call() {
         let expected = indoc! {"
             ; ModuleID = 'test'
             source_filename = \"test\"
+            target datalayout = \"e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128\"
+            target triple = \"x86_64-unknown-linux-gnu\"
 
             ; Function Attrs: mustprogress nofree norecurse nosync nounwind readnone willreturn
             define i32 @testing(i32 %0) local_unnamed_addr #0 !dbg !5 {
@@ -427,7 +452,7 @@ fn test_classes_list_type_new() {
     let llvm_usize = generator.get_size_type(&ctx);
 
     let llvm_list = ListType::new(&generator, &ctx, llvm_i32.into());
-    assert!(ListType::is_type(llvm_list.as_base_type(), llvm_usize).is_ok());
+    assert!(ListType::is_representable(llvm_list.as_base_type(), llvm_usize).is_ok());
 }
 
 #[test]
@@ -435,7 +460,7 @@ fn test_classes_range_type_new() {
     let ctx = inkwell::context::Context::create();
 
     let llvm_range = RangeType::new(&ctx);
-    assert!(RangeType::is_type(llvm_range.as_base_type()).is_ok());
+    assert!(RangeType::is_representable(llvm_range.as_base_type()).is_ok());
 }
 
 #[test]
@@ -447,5 +472,5 @@ fn test_classes_ndarray_type_new() {
     let llvm_usize = generator.get_size_type(&ctx);
 
     let llvm_ndarray = NDArrayType::new(&generator, &ctx, llvm_i32.into());
-    assert!(NDArrayType::is_type(llvm_ndarray.as_base_type(), llvm_usize).is_ok());
+    assert!(NDArrayType::is_representable(llvm_ndarray.as_base_type(), llvm_usize).is_ok());
 }

nac3core/src/codegen/types/list.rs

@@ -0,0 +1,192 @@
+use inkwell::{
+    context::Context,
+    types::{AnyTypeEnum, BasicType, BasicTypeEnum, IntType, PointerType},
+    values::IntValue,
+    AddressSpace,
+};
+
+use super::ProxyType;
+use crate::codegen::{
+    values::{ArraySliceValue, ListValue, ProxyValue},
+    CodeGenContext, CodeGenerator,
+};
+
+/// Proxy type for a `list` type in LLVM.
+#[derive(Debug, PartialEq, Eq, Clone, Copy)]
+pub struct ListType<'ctx> {
+    ty: PointerType<'ctx>,
+    llvm_usize: IntType<'ctx>,
+}
+
+impl<'ctx> ListType<'ctx> {
+    /// Checks whether `llvm_ty` represents a `list` type, returning [Err] if it does not.
+    pub fn is_representable(
+        llvm_ty: PointerType<'ctx>,
+        llvm_usize: IntType<'ctx>,
+    ) -> Result<(), String> {
+        let llvm_list_ty = llvm_ty.get_element_type();
+        let AnyTypeEnum::StructType(llvm_list_ty) = llvm_list_ty else {
+            return Err(format!("Expected struct type for `list` type, got {llvm_list_ty}"));
+        };
+        if llvm_list_ty.count_fields() != 2 {
+            return Err(format!(
+                "Expected 2 fields in `list`, got {}",
+                llvm_list_ty.count_fields()
+            ));
+        }
+
+        let list_size_ty = llvm_list_ty.get_field_type_at_index(0).unwrap();
+        let Ok(_) = PointerType::try_from(list_size_ty) else {
+            return Err(format!("Expected pointer type for `list.0`, got {list_size_ty}"));
+        };
+
+        let list_data_ty = llvm_list_ty.get_field_type_at_index(1).unwrap();
+        let Ok(list_data_ty) = IntType::try_from(list_data_ty) else {
+            return Err(format!("Expected int type for `list.1`, got {list_data_ty}"));
+        };
+        if list_data_ty.get_bit_width() != llvm_usize.get_bit_width() {
+            return Err(format!(
+                "Expected {}-bit int type for `list.1`, got {}-bit int",
+                llvm_usize.get_bit_width(),
+                list_data_ty.get_bit_width()
+            ));
+        }
+
+        Ok(())
+    }
+
+    /// Creates an LLVM type corresponding to the expected structure of a `List`.
+    #[must_use]
+    fn llvm_type(
+        ctx: &'ctx Context,
+        element_type: BasicTypeEnum<'ctx>,
+        llvm_usize: IntType<'ctx>,
+    ) -> PointerType<'ctx> {
+        // struct List { data: T*, size: size_t }
+        let field_tys = [element_type.ptr_type(AddressSpace::default()).into(), llvm_usize.into()];
+
+        ctx.struct_type(&field_tys, false).ptr_type(AddressSpace::default())
+    }
+
+    /// Creates an instance of [`ListType`].
+    #[must_use]
+    pub fn new<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        element_type: BasicTypeEnum<'ctx>,
+    ) -> Self {
+        let llvm_usize = generator.get_size_type(ctx);
+        let llvm_list = Self::llvm_type(ctx, element_type, llvm_usize);
+
+        ListType::from_type(llvm_list, llvm_usize)
+    }
+
+    /// Creates an [`ListType`] from a [`PointerType`].
+    #[must_use]
+    pub fn from_type(ptr_ty: PointerType<'ctx>, llvm_usize: IntType<'ctx>) -> Self {
+        debug_assert!(Self::is_representable(ptr_ty, llvm_usize).is_ok());
+
+        ListType { ty: ptr_ty, llvm_usize }
+    }
+
+    /// Returns the type of the `size` field of this `list` type.
+    #[must_use]
+    pub fn size_type(&self) -> IntType<'ctx> {
+        self.as_base_type()
+            .get_element_type()
+            .into_struct_type()
+            .get_field_type_at_index(1)
+            .map(BasicTypeEnum::into_int_type)
+            .unwrap()
+    }
+
+    /// Returns the element type of this `list` type.
+    #[must_use]
+    pub fn element_type(&self) -> AnyTypeEnum<'ctx> {
+        self.as_base_type()
+            .get_element_type()
+            .into_struct_type()
+            .get_field_type_at_index(0)
+            .map(BasicTypeEnum::into_pointer_type)
+            .map(PointerType::get_element_type)
+            .unwrap()
+    }
+}
+
+impl<'ctx> ProxyType<'ctx> for ListType<'ctx> {
+    type Base = PointerType<'ctx>;
+    type Value = ListValue<'ctx>;
+
+    fn is_type<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        llvm_ty: impl BasicType<'ctx>,
+    ) -> Result<(), String> {
+        if let BasicTypeEnum::PointerType(ty) = llvm_ty.as_basic_type_enum() {
+            <Self as ProxyType<'ctx>>::is_representable(generator, ctx, ty)
+        } else {
+            Err(format!("Expected pointer type, got {llvm_ty:?}"))
+        }
+    }
+
+    fn is_representable<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        llvm_ty: Self::Base,
+    ) -> Result<(), String> {
+        Self::is_representable(llvm_ty, generator.get_size_type(ctx))
+    }
+
+    fn new_value<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        name: Option<&'ctx str>,
+    ) -> Self::Value {
+        self.map_value(
+            generator
+                .gen_var_alloc(
+                    ctx,
+                    self.as_base_type().get_element_type().into_struct_type().into(),
+                    name,
+                )
+                .unwrap(),
+            name,
+        )
+    }
+
+    fn new_array_value<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        size: IntValue<'ctx>,
+        name: Option<&'ctx str>,
+    ) -> ArraySliceValue<'ctx> {
+        generator
+            .gen_array_var_alloc(
+                ctx,
+                self.as_base_type().get_element_type().into_struct_type().into(),
+                size,
+                name,
+            )
+            .unwrap()
+    }
+
+    fn map_value(
+        &self,
+        value: <Self::Value as ProxyValue<'ctx>>::Base,
+        name: Option<&'ctx str>,
+    ) -> Self::Value {
+        Self::Value::from_pointer_value(value, self.llvm_usize, name)
+    }
+
+    fn as_base_type(&self) -> Self::Base {
+        self.ty
+    }
+}
+
+impl<'ctx> From<ListType<'ctx>> for PointerType<'ctx> {
+    fn from(value: ListType<'ctx>) -> Self {
+        value.as_base_type()
+    }
+}

nac3core/src/codegen/types/mod.rs

@@ -0,0 +1,64 @@
+use inkwell::{context::Context, types::BasicType, values::IntValue};
+
+use super::{
+    values::{ArraySliceValue, ProxyValue},
+    {CodeGenContext, CodeGenerator},
+};
+pub use list::*;
+pub use ndarray::*;
+pub use range::*;
+
+mod list;
+mod ndarray;
+mod range;
+pub mod structure;
+
+/// A LLVM type that is used to represent a corresponding type in NAC3.
+pub trait ProxyType<'ctx>: Into<Self::Base> {
+    /// The LLVM type of which values of this type possess. This is usually a
+    /// [LLVM pointer type][PointerType] for any non-primitive types.
+    type Base: BasicType<'ctx>;
+
+    /// The type of values represented by this type.
+    type Value: ProxyValue<'ctx, Type = Self>;
+
+    fn is_type<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        llvm_ty: impl BasicType<'ctx>,
+    ) -> Result<(), String>;
+
+    /// Checks whether `llvm_ty` can be represented by this [`ProxyType`].
+    fn is_representable<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        llvm_ty: Self::Base,
+    ) -> Result<(), String>;
+
+    /// Creates a new value of this type.
+    fn new_value<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        name: Option<&'ctx str>,
+    ) -> Self::Value;
+
+    /// Creates a new array value of this type.
+    fn new_array_value<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        size: IntValue<'ctx>,
+        name: Option<&'ctx str>,
+    ) -> ArraySliceValue<'ctx>;
+
+    /// Converts an existing value into a [`ProxyValue`] of this type.
+    fn map_value(
+        &self,
+        value: <Self::Value as ProxyValue<'ctx>>::Base,
+        name: Option<&'ctx str>,
+    ) -> Self::Value;
+
+    /// Returns the [base type][Self::Base] of this proxy.
+    fn as_base_type(&self) -> Self::Base;
+}

nac3core/src/codegen/types/ndarray.rs

@@ -0,0 +1,258 @@
+use inkwell::{
+    context::Context,
+    types::{AnyTypeEnum, BasicType, BasicTypeEnum, IntType, PointerType},
+    values::{IntValue, PointerValue},
+    AddressSpace,
+};
+use itertools::Itertools;
+
+use nac3core_derive::StructFields;
+
+use super::{
+    structure::{StructField, StructFields},
+    ProxyType,
+};
+use crate::codegen::{
+    values::{ArraySliceValue, NDArrayValue, ProxyValue},
+    {CodeGenContext, CodeGenerator},
+};
+
+/// Proxy type for a `ndarray` type in LLVM.
+#[derive(Debug, PartialEq, Eq, Clone, Copy)]
+pub struct NDArrayType<'ctx> {
+    ty: PointerType<'ctx>,
+    dtype: BasicTypeEnum<'ctx>,
+    llvm_usize: IntType<'ctx>,
+}
+
+#[derive(PartialEq, Eq, Clone, Copy, StructFields)]
+pub struct NDArrayStructFields<'ctx> {
+    #[value_type(usize)]
+    pub ndims: StructField<'ctx, IntValue<'ctx>>,
+    #[value_type(usize.ptr_type(AddressSpace::default()))]
+    pub shape: StructField<'ctx, PointerValue<'ctx>>,
+    #[value_type(i8_type().ptr_type(AddressSpace::default()))]
+    pub data: StructField<'ctx, PointerValue<'ctx>>,
+}
+
+impl<'ctx> NDArrayType<'ctx> {
+    /// Checks whether `llvm_ty` represents a `ndarray` type, returning [Err] if it does not.
+    pub fn is_representable(
+        llvm_ty: PointerType<'ctx>,
+        llvm_usize: IntType<'ctx>,
+    ) -> Result<(), String> {
+        let llvm_ndarray_ty = llvm_ty.get_element_type();
+        let AnyTypeEnum::StructType(llvm_ndarray_ty) = llvm_ndarray_ty else {
+            return Err(format!("Expected struct type for `NDArray` type, got {llvm_ndarray_ty}"));
+        };
+        if llvm_ndarray_ty.count_fields() != 3 {
+            return Err(format!(
+                "Expected 3 fields in `NDArray`, got {}",
+                llvm_ndarray_ty.count_fields()
+            ));
+        }
+
+        let ndarray_ndims_ty = llvm_ndarray_ty.get_field_type_at_index(0).unwrap();
+        let Ok(ndarray_ndims_ty) = IntType::try_from(ndarray_ndims_ty) else {
+            return Err(format!("Expected int type for `ndarray.0`, got {ndarray_ndims_ty}"));
+        };
+        if ndarray_ndims_ty.get_bit_width() != llvm_usize.get_bit_width() {
+            return Err(format!(
+                "Expected {}-bit int type for `ndarray.0`, got {}-bit int",
+                llvm_usize.get_bit_width(),
+                ndarray_ndims_ty.get_bit_width()
+            ));
+        }
+
+        let ndarray_dims_ty = llvm_ndarray_ty.get_field_type_at_index(1).unwrap();
+        let Ok(ndarray_pdims) = PointerType::try_from(ndarray_dims_ty) else {
+            return Err(format!("Expected pointer type for `ndarray.1`, got {ndarray_dims_ty}"));
+        };
+        let ndarray_dims = ndarray_pdims.get_element_type();
+        let Ok(ndarray_dims) = IntType::try_from(ndarray_dims) else {
+            return Err(format!(
+                "Expected pointer-to-int type for `ndarray.1`, got pointer-to-{ndarray_dims}"
+            ));
+        };
+        if ndarray_dims.get_bit_width() != llvm_usize.get_bit_width() {
+            return Err(format!(
+                "Expected pointer-to-{}-bit int type for `ndarray.1`, got pointer-to-{}-bit int",
+                llvm_usize.get_bit_width(),
+                ndarray_dims.get_bit_width()
+            ));
+        }
+
+        let ndarray_data_ty = llvm_ndarray_ty.get_field_type_at_index(2).unwrap();
+        let Ok(ndarray_pdata) = PointerType::try_from(ndarray_data_ty) else {
+            return Err(format!("Expected pointer type for `ndarray.2`, got {ndarray_data_ty}"));
+        };
+        let ndarray_data = ndarray_pdata.get_element_type();
+        let Ok(ndarray_data) = IntType::try_from(ndarray_data) else {
+            return Err(format!(
+                "Expected pointer-to-int type for `ndarray.2`, got pointer-to-{ndarray_data}"
+            ));
+        };
+        if ndarray_data.get_bit_width() != 8 {
+            return Err(format!(
+                "Expected pointer-to-8-bit int type for `ndarray.1`, got pointer-to-{}-bit int",
+                ndarray_data.get_bit_width()
+            ));
+        }
+
+        Ok(())
+    }
+
+    // TODO: Move this into e.g. StructProxyType
+    #[must_use]
+    fn fields(ctx: &'ctx Context, llvm_usize: IntType<'ctx>) -> NDArrayStructFields<'ctx> {
+        NDArrayStructFields::new(ctx, llvm_usize)
+    }
+
+    // TODO: Move this into e.g. StructProxyType
+    #[must_use]
+    pub fn get_fields(
+        &self,
+        ctx: &'ctx Context,
+        llvm_usize: IntType<'ctx>,
+    ) -> NDArrayStructFields<'ctx> {
+        Self::fields(ctx, llvm_usize)
+    }
+
+    /// Creates an LLVM type corresponding to the expected structure of an `NDArray`.
+    #[must_use]
+    fn llvm_type(ctx: &'ctx Context, llvm_usize: IntType<'ctx>) -> PointerType<'ctx> {
+        // struct NDArray { num_dims: size_t, dims: size_t*, data: i8* }
+        //
+        // * data    : Pointer to an array containing the array data
+        // * itemsize: The size of each NDArray elements in bytes
+        // * ndims   : Number of dimensions in the array
+        // * shape   : Pointer to an array containing the shape of the NDArray
+        // * strides : Pointer to an array indicating the number of bytes between each element at a dimension
+        let field_tys =
+            Self::fields(ctx, llvm_usize).into_iter().map(|field| field.1).collect_vec();
+
+        ctx.struct_type(&field_tys, false).ptr_type(AddressSpace::default())
+    }
+
+    /// Creates an instance of [`NDArrayType`].
+    #[must_use]
+    pub fn new<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        dtype: BasicTypeEnum<'ctx>,
+    ) -> Self {
+        let llvm_usize = generator.get_size_type(ctx);
+        let llvm_ndarray = Self::llvm_type(ctx, llvm_usize);
+
+        NDArrayType { ty: llvm_ndarray, dtype, llvm_usize }
+    }
+
+    /// Creates an [`NDArrayType`] from a [`PointerType`] representing an `NDArray`.
+    #[must_use]
+    pub fn from_type(
+        ptr_ty: PointerType<'ctx>,
+        dtype: BasicTypeEnum<'ctx>,
+        llvm_usize: IntType<'ctx>,
+    ) -> Self {
+        debug_assert!(Self::is_representable(ptr_ty, llvm_usize).is_ok());
+
+        NDArrayType { ty: ptr_ty, dtype, llvm_usize }
+    }
+
+    /// Returns the type of the `size` field of this `ndarray` type.
+    #[must_use]
+    pub fn size_type(&self) -> IntType<'ctx> {
+        self.as_base_type()
+            .get_element_type()
+            .into_struct_type()
+            .get_field_type_at_index(0)
+            .map(BasicTypeEnum::into_int_type)
+            .unwrap()
+    }
+
+    /// Returns the element type of this `ndarray` type.
+    #[must_use]
+    pub fn element_type(&self) -> BasicTypeEnum<'ctx> {
+        self.dtype
+    }
+}
+
+impl<'ctx> ProxyType<'ctx> for NDArrayType<'ctx> {
+    type Base = PointerType<'ctx>;
+    type Value = NDArrayValue<'ctx>;
+
+    fn is_type<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        llvm_ty: impl BasicType<'ctx>,
+    ) -> Result<(), String> {
+        if let BasicTypeEnum::PointerType(ty) = llvm_ty.as_basic_type_enum() {
+            <Self as ProxyType<'ctx>>::is_representable(generator, ctx, ty)
+        } else {
+            Err(format!("Expected pointer type, got {llvm_ty:?}"))
+        }
+    }
+
+    fn is_representable<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        llvm_ty: Self::Base,
+    ) -> Result<(), String> {
+        Self::is_representable(llvm_ty, generator.get_size_type(ctx))
+    }
+
+    fn new_value<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        name: Option<&'ctx str>,
+    ) -> Self::Value {
+        self.map_value(
+            generator
+                .gen_var_alloc(
+                    ctx,
+                    self.as_base_type().get_element_type().into_struct_type().into(),
+                    name,
+                )
+                .unwrap(),
+            name,
+        )
+    }
+
+    fn new_array_value<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        size: IntValue<'ctx>,
+        name: Option<&'ctx str>,
+    ) -> ArraySliceValue<'ctx> {
+        generator
+            .gen_array_var_alloc(
+                ctx,
+                self.as_base_type().get_element_type().into_struct_type().into(),
+                size,
+                name,
+            )
+            .unwrap()
+    }
+
+    fn map_value(
+        &self,
+        value: <Self::Value as ProxyValue<'ctx>>::Base,
+        name: Option<&'ctx str>,
+    ) -> Self::Value {
+        debug_assert_eq!(value.get_type(), self.as_base_type());
+
+        NDArrayValue::from_pointer_value(value, self.dtype, self.llvm_usize, name)
+    }
+
+    fn as_base_type(&self) -> Self::Base {
+        self.ty
+    }
+}
+
+impl<'ctx> From<NDArrayType<'ctx>> for PointerType<'ctx> {
+    fn from(value: NDArrayType<'ctx>) -> Self {
+        value.as_base_type()
+    }
+}

nac3core/src/codegen/types/range.rs

@@ -0,0 +1,159 @@
+use inkwell::{
+    context::Context,
+    types::{AnyTypeEnum, BasicType, BasicTypeEnum, IntType, PointerType},
+    values::IntValue,
+    AddressSpace,
+};
+
+use super::ProxyType;
+use crate::codegen::{
+    values::{ArraySliceValue, ProxyValue, RangeValue},
+    {CodeGenContext, CodeGenerator},
+};
+
+/// Proxy type for a `range` type in LLVM.
+#[derive(Debug, PartialEq, Eq, Clone, Copy)]
+pub struct RangeType<'ctx> {
+    ty: PointerType<'ctx>,
+}
+
+impl<'ctx> RangeType<'ctx> {
+    /// Checks whether `llvm_ty` represents a `range` type, returning [Err] if it does not.
+    pub fn is_representable(llvm_ty: PointerType<'ctx>) -> Result<(), String> {
+        let llvm_range_ty = llvm_ty.get_element_type();
+        let AnyTypeEnum::ArrayType(llvm_range_ty) = llvm_range_ty else {
+            return Err(format!("Expected array type for `range` type, got {llvm_range_ty}"));
+        };
+        if llvm_range_ty.len() != 3 {
+            return Err(format!(
+                "Expected 3 elements for `range` type, got {}",
+                llvm_range_ty.len()
+            ));
+        }
+
+        let llvm_range_elem_ty = llvm_range_ty.get_element_type();
+        let Ok(llvm_range_elem_ty) = IntType::try_from(llvm_range_elem_ty) else {
+            return Err(format!(
+                "Expected int type for `range` element type, got {llvm_range_elem_ty}"
+            ));
+        };
+        if llvm_range_elem_ty.get_bit_width() != 32 {
+            return Err(format!(
+                "Expected 32-bit int type for `range` element type, got {}",
+                llvm_range_elem_ty.get_bit_width()
+            ));
+        }
+
+        Ok(())
+    }
+
+    /// Creates an LLVM type corresponding to the expected structure of a `Range`.
+    #[must_use]
+    fn llvm_type(ctx: &'ctx Context) -> PointerType<'ctx> {
+        // typedef int32_t Range[3];
+        let llvm_i32 = ctx.i32_type();
+        llvm_i32.array_type(3).ptr_type(AddressSpace::default())
+    }
+
+    /// Creates an instance of [`RangeType`].
+    #[must_use]
+    pub fn new(ctx: &'ctx Context) -> Self {
+        let llvm_range = Self::llvm_type(ctx);
+
+        RangeType::from_type(llvm_range)
+    }
+
+    /// Creates an [`RangeType`] from a [`PointerType`].
+    #[must_use]
+    pub fn from_type(ptr_ty: PointerType<'ctx>) -> Self {
+        debug_assert!(Self::is_representable(ptr_ty).is_ok());
+
+        RangeType { ty: ptr_ty }
+    }
+
+    /// Returns the type of all fields of this `range` type.
+    #[must_use]
+    pub fn value_type(&self) -> IntType<'ctx> {
+        self.as_base_type().get_element_type().into_array_type().get_element_type().into_int_type()
+    }
+}
+
+impl<'ctx> ProxyType<'ctx> for RangeType<'ctx> {
+    type Base = PointerType<'ctx>;
+    type Value = RangeValue<'ctx>;
+
+    fn is_type<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        llvm_ty: impl BasicType<'ctx>,
+    ) -> Result<(), String> {
+        if let BasicTypeEnum::PointerType(ty) = llvm_ty.as_basic_type_enum() {
+            <Self as ProxyType<'ctx>>::is_representable(generator, ctx, ty)
+        } else {
+            Err(format!("Expected pointer type, got {llvm_ty:?}"))
+        }
+    }
+
+    fn is_representable<G: CodeGenerator + ?Sized>(
+        _: &G,
+        _: &'ctx Context,
+        llvm_ty: Self::Base,
+    ) -> Result<(), String> {
+        Self::is_representable(llvm_ty)
+    }
+
+    fn new_value<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        name: Option<&'ctx str>,
+    ) -> Self::Value {
+        self.map_value(
+            generator
+                .gen_var_alloc(
+                    ctx,
+                    self.as_base_type().get_element_type().into_struct_type().into(),
+                    name,
+                )
+                .unwrap(),
+            name,
+        )
+    }
+
+    fn new_array_value<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        size: IntValue<'ctx>,
+        name: Option<&'ctx str>,
+    ) -> ArraySliceValue<'ctx> {
+        generator
+            .gen_array_var_alloc(
+                ctx,
+                self.as_base_type().get_element_type().into_struct_type().into(),
+                size,
+                name,
+            )
+            .unwrap()
+    }
+
+    fn map_value(
+        &self,
+        value: <Self::Value as ProxyValue<'ctx>>::Base,
+        name: Option<&'ctx str>,
+    ) -> Self::Value {
+        debug_assert_eq!(value.get_type(), self.as_base_type());
+
+        RangeValue::from_pointer_value(value, name)
+    }
+
+    fn as_base_type(&self) -> Self::Base {
+        self.ty
+    }
+}
+
+impl<'ctx> From<RangeType<'ctx>> for PointerType<'ctx> {
+    fn from(value: RangeType<'ctx>) -> Self {
+        value.as_base_type()
+    }
+}

nac3core/src/codegen/types/structure.rs

@@ -0,0 +1,203 @@
+use std::marker::PhantomData;
+
+use inkwell::{
+    context::AsContextRef,
+    types::{BasicTypeEnum, IntType},
+    values::{BasicValue, BasicValueEnum, IntValue, PointerValue, StructValue},
+};
+
+use crate::codegen::CodeGenContext;
+
+/// Trait indicating that the structure is a field-wise representation of an LLVM structure.
+///
+/// # Usage
+///
+/// For example, for a simple C-slice LLVM structure:
+///
+/// ```ignore
+/// struct CSliceFields<'ctx> {
+///     ptr: StructField<'ctx, PointerValue<'ctx>>,
+///     len: StructField<'ctx, IntValue<'ctx>>
+/// }
+/// ```
+pub trait StructFields<'ctx>: Eq + Copy {
+    /// Creates an instance of [`StructFields`] using the given `ctx` and `size_t` types.
+    fn new(ctx: impl AsContextRef<'ctx>, llvm_usize: IntType<'ctx>) -> Self;
+
+    /// Returns a [`Vec`] that contains the fields of the structure in the order as they appear in
+    /// the type definition.
+    #[must_use]
+    fn to_vec(&self) -> Vec<(&'static str, BasicTypeEnum<'ctx>)>;
+
+    /// Returns a [`Iterator`] that contains the fields of the structure in the order as they appear
+    /// in the type definition.
+    #[must_use]
+    fn iter(&self) -> impl Iterator<Item = (&'static str, BasicTypeEnum<'ctx>)> {
+        self.to_vec().into_iter()
+    }
+
+    /// Returns a [`Vec`] that contains the fields of the structure in the order as they appear in
+    /// the type definition.
+    #[must_use]
+    fn into_vec(self) -> Vec<(&'static str, BasicTypeEnum<'ctx>)>
+    where
+        Self: Sized,
+    {
+        self.to_vec()
+    }
+
+    /// Returns a [`Iterator`] that contains the fields of the structure in the order as they appear
+    /// in the type definition.
+    #[must_use]
+    fn into_iter(self) -> impl Iterator<Item = (&'static str, BasicTypeEnum<'ctx>)>
+    where
+        Self: Sized,
+    {
+        self.into_vec().into_iter()
+    }
+}
+
+/// A single field of an LLVM structure.
+#[derive(Debug, PartialEq, Eq, Clone, Copy)]
+pub struct StructField<'ctx, Value>
+where
+    Value: BasicValue<'ctx> + TryFrom<BasicValueEnum<'ctx>, Error = ()>,
+{
+    /// The index of this field within the structure.
+    index: u32,
+
+    /// The name of this field.
+    name: &'static str,
+
+    /// The type of this field.
+    ty: BasicTypeEnum<'ctx>,
+
+    /// Instance of [`PhantomData`] containing [`Value`], used to implement automatic downcasts.
+    _value_ty: PhantomData<Value>,
+}
+
+impl<'ctx, Value> StructField<'ctx, Value>
+where
+    Value: BasicValue<'ctx> + TryFrom<BasicValueEnum<'ctx>, Error = ()>,
+{
+    /// Creates an instance of [`StructField`].
+    ///
+    /// * `idx_counter` - The instance of [`FieldIndexCounter`] used to track the current field
+    ///   index.
+    /// * `name` - Name of the field.
+    /// * `ty` - The type of this field.
+    pub fn create(
+        idx_counter: &mut FieldIndexCounter,
+        name: &'static str,
+        ty: impl Into<BasicTypeEnum<'ctx>>,
+    ) -> Self {
+        StructField { index: idx_counter.increment(), name, ty: ty.into(), _value_ty: PhantomData }
+    }
+
+    /// Creates an instance of [`StructField`] with a given index.
+    ///
+    /// * `index` - The index of this field within its enclosing structure.
+    /// * `name` - Name of the field.
+    /// * `ty` - The type of this field.
+    pub fn create_at(index: u32, name: &'static str, ty: impl Into<BasicTypeEnum<'ctx>>) -> Self {
+        StructField { index, name, ty: ty.into(), _value_ty: PhantomData }
+    }
+
+    /// Creates a pointer to this field in an arbitrary structure by performing a `getelementptr i32
+    /// {idx...}, i32 {self.index}`.
+    pub fn ptr_by_array_gep(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        pobj: PointerValue<'ctx>,
+        idx: &[IntValue<'ctx>],
+    ) -> PointerValue<'ctx> {
+        unsafe {
+            ctx.builder.build_in_bounds_gep(
+                pobj,
+                &[idx, &[ctx.ctx.i32_type().const_int(u64::from(self.index), false)]].concat(),
+                "",
+            )
+        }
+        .unwrap()
+    }
+
+    /// Creates a pointer to this field in an arbitrary structure by performing the equivalent of
+    /// `getelementptr i32 0, i32 {self.index}`.
+    pub fn ptr_by_gep(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        pobj: PointerValue<'ctx>,
+        obj_name: Option<&'ctx str>,
+    ) -> PointerValue<'ctx> {
+        ctx.builder
+            .build_struct_gep(
+                pobj,
+                self.index,
+                &obj_name.map(|name| format!("{name}.{}.addr", self.name)).unwrap_or_default(),
+            )
+            .unwrap()
+    }
+
+    /// Gets the value of this field for a given `obj`.
+    #[must_use]
+    pub fn get_from_value(&self, obj: StructValue<'ctx>) -> Value {
+        obj.get_field_at_index(self.index).and_then(|value| Value::try_from(value).ok()).unwrap()
+    }
+
+    /// Sets the value of this field for a given `obj`.
+    pub fn set_from_value(&self, obj: StructValue<'ctx>, value: Value) {
+        obj.set_field_at_index(self.index, value);
+    }
+
+    /// Gets the value of this field for a pointer-to-structure.
+    pub fn get(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        pobj: PointerValue<'ctx>,
+        obj_name: Option<&'ctx str>,
+    ) -> Value {
+        ctx.builder
+            .build_load(
+                self.ptr_by_gep(ctx, pobj, obj_name),
+                &obj_name.map(|name| format!("{name}.{}", self.name)).unwrap_or_default(),
+            )
+            .map_err(|_| ())
+            .and_then(|value| Value::try_from(value))
+            .unwrap()
+    }
+
+    /// Sets the value of this field for a pointer-to-structure.
+    pub fn set(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        pobj: PointerValue<'ctx>,
+        value: Value,
+        obj_name: Option<&'ctx str>,
+    ) {
+        ctx.builder.build_store(self.ptr_by_gep(ctx, pobj, obj_name), value).unwrap();
+    }
+}
+
+impl<'ctx, Value> From<StructField<'ctx, Value>> for (&'static str, BasicTypeEnum<'ctx>)
+where
+    Value: BasicValue<'ctx> + TryFrom<BasicValueEnum<'ctx>, Error = ()>,
+{
+    fn from(value: StructField<'ctx, Value>) -> Self {
+        (value.name, value.ty)
+    }
+}
+
+/// A counter that tracks the next index of a field using a monotonically increasing counter.
+#[derive(Default, Debug, PartialEq, Eq, Clone, Copy)]
+pub struct FieldIndexCounter(u32);
+
+impl FieldIndexCounter {
+    /// Increments the number stored by this counter, returning the previous value.
+    ///
+    /// Functionally equivalent to `i++` in C-based languages.
+    pub fn increment(&mut self) -> u32 {
+        let v = self.0;
+        self.0 += 1;
+        v
+    }
+}

nac3core/src/codegen/values/array.rs

@@ -0,0 +1,426 @@
+use inkwell::{
+    types::AnyTypeEnum,
+    values::{BasicValueEnum, IntValue, PointerValue},
+    IntPredicate,
+};
+
+use crate::codegen::{CodeGenContext, CodeGenerator};
+
+/// An LLVM value that is array-like, i.e. it contains a contiguous, sequenced collection of
+/// elements.
+pub trait ArrayLikeValue<'ctx> {
+    /// Returns the element type of this array-like value.
+    fn element_type<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+    ) -> AnyTypeEnum<'ctx>;
+
+    /// Returns the base pointer to the array.
+    fn base_ptr<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+    ) -> PointerValue<'ctx>;
+
+    /// Returns the size of this array-like value.
+    fn size<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+    ) -> IntValue<'ctx>;
+
+    /// Returns a [`ArraySliceValue`] representing this value.
+    fn as_slice_value<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+    ) -> ArraySliceValue<'ctx> {
+        ArraySliceValue::from_ptr_val(
+            self.base_ptr(ctx, generator),
+            self.size(ctx, generator),
+            None,
+        )
+    }
+}
+
+/// An array-like value that can be indexed by memory offset.
+pub trait ArrayLikeIndexer<'ctx, Index = IntValue<'ctx>>: ArrayLikeValue<'ctx> {
+    /// # Safety
+    ///
+    /// This function should be called with a valid index.
+    unsafe fn ptr_offset_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx>;
+
+    /// Returns the pointer to the data at the `idx`-th index.
+    fn ptr_offset<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx>;
+}
+
+/// An array-like value that can have its array elements accessed as a [`BasicValueEnum`].
+pub trait UntypedArrayLikeAccessor<'ctx, Index = IntValue<'ctx>>:
+    ArrayLikeIndexer<'ctx, Index>
+{
+    /// # Safety
+    ///
+    /// This function should be called with a valid index.
+    unsafe fn get_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        name: Option<&str>,
+    ) -> BasicValueEnum<'ctx> {
+        let ptr = unsafe { self.ptr_offset_unchecked(ctx, generator, idx, name) };
+        ctx.builder.build_load(ptr, name.unwrap_or_default()).unwrap()
+    }
+
+    /// Returns the data at the `idx`-th index.
+    fn get<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        name: Option<&str>,
+    ) -> BasicValueEnum<'ctx> {
+        let ptr = self.ptr_offset(ctx, generator, idx, name);
+        ctx.builder.build_load(ptr, name.unwrap_or_default()).unwrap()
+    }
+}
+
+/// An array-like value that can have its array elements mutated as a [`BasicValueEnum`].
+pub trait UntypedArrayLikeMutator<'ctx, Index = IntValue<'ctx>>:
+    ArrayLikeIndexer<'ctx, Index>
+{
+    /// # Safety
+    ///
+    /// This function should be called with a valid index.
+    unsafe fn set_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        value: BasicValueEnum<'ctx>,
+    ) {
+        let ptr = unsafe { self.ptr_offset_unchecked(ctx, generator, idx, None) };
+        ctx.builder.build_store(ptr, value).unwrap();
+    }
+
+    /// Sets the data at the `idx`-th index.
+    fn set<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        value: BasicValueEnum<'ctx>,
+    ) {
+        let ptr = self.ptr_offset(ctx, generator, idx, None);
+        ctx.builder.build_store(ptr, value).unwrap();
+    }
+}
+
+/// An array-like value that can have its array elements accessed as an arbitrary type `T`.
+pub trait TypedArrayLikeAccessor<'ctx, T, Index = IntValue<'ctx>>:
+    UntypedArrayLikeAccessor<'ctx, Index>
+{
+    /// Casts an element from [`BasicValueEnum`] into `T`.
+    fn downcast_to_type(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        value: BasicValueEnum<'ctx>,
+    ) -> T;
+
+    /// # Safety
+    ///
+    /// This function should be called with a valid index.
+    unsafe fn get_typed_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        name: Option<&str>,
+    ) -> T {
+        let value = unsafe { self.get_unchecked(ctx, generator, idx, name) };
+        self.downcast_to_type(ctx, value)
+    }
+
+    /// Returns the data at the `idx`-th index.
+    fn get_typed<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        name: Option<&str>,
+    ) -> T {
+        let value = self.get(ctx, generator, idx, name);
+        self.downcast_to_type(ctx, value)
+    }
+}
+
+/// An array-like value that can have its array elements mutated as an arbitrary type `T`.
+pub trait TypedArrayLikeMutator<'ctx, T, Index = IntValue<'ctx>>:
+    UntypedArrayLikeMutator<'ctx, Index>
+{
+    /// Casts an element from T into [`BasicValueEnum`].
+    fn upcast_from_type(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        value: T,
+    ) -> BasicValueEnum<'ctx>;
+
+    /// # Safety
+    ///
+    /// This function should be called with a valid index.
+    unsafe fn set_typed_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        value: T,
+    ) {
+        let value = self.upcast_from_type(ctx, value);
+        unsafe { self.set_unchecked(ctx, generator, idx, value) }
+    }
+
+    /// Sets the data at the `idx`-th index.
+    fn set_typed<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        value: T,
+    ) {
+        let value = self.upcast_from_type(ctx, value);
+        self.set(ctx, generator, idx, value);
+    }
+}
+
+/// Type alias for a function that casts a [`BasicValueEnum`] into a `T`.
+type ValueDowncastFn<'ctx, T> =
+    Box<dyn Fn(&mut CodeGenContext<'ctx, '_>, BasicValueEnum<'ctx>) -> T>;
+/// Type alias for a function that casts a `T` into a [`BasicValueEnum`].
+type ValueUpcastFn<'ctx, T> = Box<dyn Fn(&mut CodeGenContext<'ctx, '_>, T) -> BasicValueEnum<'ctx>>;
+
+/// An adapter for constraining untyped array values as typed values.
+pub struct TypedArrayLikeAdapter<'ctx, T, Adapted: ArrayLikeValue<'ctx> = ArraySliceValue<'ctx>> {
+    adapted: Adapted,
+    downcast_fn: ValueDowncastFn<'ctx, T>,
+    upcast_fn: ValueUpcastFn<'ctx, T>,
+}
+
+impl<'ctx, T, Adapted> TypedArrayLikeAdapter<'ctx, T, Adapted>
+where
+    Adapted: ArrayLikeValue<'ctx>,
+{
+    /// Creates a [`TypedArrayLikeAdapter`].
+    ///
+    /// * `adapted` - The value to be adapted.
+    /// * `downcast_fn` - The function converting a [`BasicValueEnum`] into a `T`.
+    /// * `upcast_fn` - The function converting a T into a [`BasicValueEnum`].
+    pub fn from(
+        adapted: Adapted,
+        downcast_fn: ValueDowncastFn<'ctx, T>,
+        upcast_fn: ValueUpcastFn<'ctx, T>,
+    ) -> Self {
+        TypedArrayLikeAdapter { adapted, downcast_fn, upcast_fn }
+    }
+}
+
+impl<'ctx, T, Adapted> ArrayLikeValue<'ctx> for TypedArrayLikeAdapter<'ctx, T, Adapted>
+where
+    Adapted: ArrayLikeValue<'ctx>,
+{
+    fn element_type<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+    ) -> AnyTypeEnum<'ctx> {
+        self.adapted.element_type(ctx, generator)
+    }
+
+    fn base_ptr<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+    ) -> PointerValue<'ctx> {
+        self.adapted.base_ptr(ctx, generator)
+    }
+
+    fn size<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+    ) -> IntValue<'ctx> {
+        self.adapted.size(ctx, generator)
+    }
+}
+
+impl<'ctx, T, Index, Adapted> ArrayLikeIndexer<'ctx, Index>
+    for TypedArrayLikeAdapter<'ctx, T, Adapted>
+where
+    Adapted: ArrayLikeIndexer<'ctx, Index>,
+{
+    unsafe fn ptr_offset_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        unsafe { self.adapted.ptr_offset_unchecked(ctx, generator, idx, name) }
+    }
+
+    fn ptr_offset<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &Index,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        self.adapted.ptr_offset(ctx, generator, idx, name)
+    }
+}
+
+impl<'ctx, T, Index, Adapted> UntypedArrayLikeAccessor<'ctx, Index>
+    for TypedArrayLikeAdapter<'ctx, T, Adapted>
+where
+    Adapted: UntypedArrayLikeAccessor<'ctx, Index>,
+{
+}
+impl<'ctx, T, Index, Adapted> UntypedArrayLikeMutator<'ctx, Index>
+    for TypedArrayLikeAdapter<'ctx, T, Adapted>
+where
+    Adapted: UntypedArrayLikeMutator<'ctx, Index>,
+{
+}
+
+impl<'ctx, T, Index, Adapted> TypedArrayLikeAccessor<'ctx, T, Index>
+    for TypedArrayLikeAdapter<'ctx, T, Adapted>
+where
+    Adapted: UntypedArrayLikeAccessor<'ctx, Index>,
+{
+    fn downcast_to_type(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        value: BasicValueEnum<'ctx>,
+    ) -> T {
+        (self.downcast_fn)(ctx, value)
+    }
+}
+
+impl<'ctx, T, Index, Adapted> TypedArrayLikeMutator<'ctx, T, Index>
+    for TypedArrayLikeAdapter<'ctx, T, Adapted>
+where
+    Adapted: UntypedArrayLikeMutator<'ctx, Index>,
+{
+    fn upcast_from_type(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        value: T,
+    ) -> BasicValueEnum<'ctx> {
+        (self.upcast_fn)(ctx, value)
+    }
+}
+
+/// An LLVM value representing an array slice, consisting of a pointer to the data and the size of
+/// the slice.
+#[derive(Copy, Clone)]
+pub struct ArraySliceValue<'ctx>(PointerValue<'ctx>, IntValue<'ctx>, Option<&'ctx str>);
+
+impl<'ctx> ArraySliceValue<'ctx> {
+    /// Creates an [`ArraySliceValue`] from a [`PointerValue`] and its size.
+    #[must_use]
+    pub fn from_ptr_val(
+        ptr: PointerValue<'ctx>,
+        size: IntValue<'ctx>,
+        name: Option<&'ctx str>,
+    ) -> Self {
+        ArraySliceValue(ptr, size, name)
+    }
+}
+
+impl<'ctx> From<ArraySliceValue<'ctx>> for PointerValue<'ctx> {
+    fn from(value: ArraySliceValue<'ctx>) -> Self {
+        value.0
+    }
+}
+
+impl<'ctx> ArrayLikeValue<'ctx> for ArraySliceValue<'ctx> {
+    fn element_type<G: CodeGenerator + ?Sized>(
+        &self,
+        _: &CodeGenContext<'ctx, '_>,
+        _: &G,
+    ) -> AnyTypeEnum<'ctx> {
+        self.0.get_type().get_element_type()
+    }
+
+    fn base_ptr<G: CodeGenerator + ?Sized>(
+        &self,
+        _: &CodeGenContext<'ctx, '_>,
+        _: &G,
+    ) -> PointerValue<'ctx> {
+        self.0
+    }
+
+    fn size<G: CodeGenerator + ?Sized>(
+        &self,
+        _: &CodeGenContext<'ctx, '_>,
+        _: &G,
+    ) -> IntValue<'ctx> {
+        self.1
+    }
+}
+
+impl<'ctx> ArrayLikeIndexer<'ctx> for ArraySliceValue<'ctx> {
+    unsafe fn ptr_offset_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &IntValue<'ctx>,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        let var_name = name.map(|v| format!("{v}.addr")).unwrap_or_default();
+
+        unsafe {
+            ctx.builder
+                .build_in_bounds_gep(self.base_ptr(ctx, generator), &[*idx], var_name.as_str())
+                .unwrap()
+        }
+    }
+
+    fn ptr_offset<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &IntValue<'ctx>,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        debug_assert_eq!(idx.get_type(), generator.get_size_type(ctx.ctx));
+
+        let size = self.size(ctx, generator);
+        let in_range = ctx.builder.build_int_compare(IntPredicate::ULT, *idx, size, "").unwrap();
+        ctx.make_assert(
+            generator,
+            in_range,
+            "0:IndexError",
+            "list index out of range",
+            [None, None, None],
+            ctx.current_loc,
+        );
+
+        unsafe { self.ptr_offset_unchecked(ctx, generator, idx, name) }
+    }
+}
+
+impl<'ctx> UntypedArrayLikeAccessor<'ctx> for ArraySliceValue<'ctx> {}
+impl<'ctx> UntypedArrayLikeMutator<'ctx> for ArraySliceValue<'ctx> {}

nac3core/src/codegen/values/list.rs

@@ -0,0 +1,241 @@
+use inkwell::{
+    types::{AnyTypeEnum, BasicType, BasicTypeEnum, IntType},
+    values::{BasicValueEnum, IntValue, PointerValue},
+    AddressSpace, IntPredicate,
+};
+
+use super::{
+    ArrayLikeIndexer, ArrayLikeValue, ProxyValue, UntypedArrayLikeAccessor, UntypedArrayLikeMutator,
+};
+use crate::codegen::{
+    types::ListType,
+    {CodeGenContext, CodeGenerator},
+};
+
+/// Proxy type for accessing a `list` value in LLVM.
+#[derive(Copy, Clone)]
+pub struct ListValue<'ctx> {
+    value: PointerValue<'ctx>,
+    llvm_usize: IntType<'ctx>,
+    name: Option<&'ctx str>,
+}
+
+impl<'ctx> ListValue<'ctx> {
+    /// Checks whether `value` is an instance of `list`, returning [Err] if `value` is not an
+    /// instance.
+    pub fn is_representable(
+        value: PointerValue<'ctx>,
+        llvm_usize: IntType<'ctx>,
+    ) -> Result<(), String> {
+        ListType::is_representable(value.get_type(), llvm_usize)
+    }
+
+    /// Creates an [`ListValue`] from a [`PointerValue`].
+    #[must_use]
+    pub fn from_pointer_value(
+        ptr: PointerValue<'ctx>,
+        llvm_usize: IntType<'ctx>,
+        name: Option<&'ctx str>,
+    ) -> Self {
+        debug_assert!(Self::is_representable(ptr, llvm_usize).is_ok());
+
+        ListValue { value: ptr, llvm_usize, name }
+    }
+
+    /// Returns the double-indirection pointer to the `data` array, as if by calling `getelementptr`
+    /// on the field.
+    fn pptr_to_data(&self, ctx: &CodeGenContext<'ctx, '_>) -> PointerValue<'ctx> {
+        let llvm_i32 = ctx.ctx.i32_type();
+        let var_name = self.name.map(|v| format!("{v}.data.addr")).unwrap_or_default();
+
+        unsafe {
+            ctx.builder
+                .build_in_bounds_gep(
+                    self.as_base_value(),
+                    &[llvm_i32.const_zero(), llvm_i32.const_zero()],
+                    var_name.as_str(),
+                )
+                .unwrap()
+        }
+    }
+
+    /// Returns the pointer to the field storing the size of this `list`.
+    fn ptr_to_size(&self, ctx: &CodeGenContext<'ctx, '_>) -> PointerValue<'ctx> {
+        let llvm_i32 = ctx.ctx.i32_type();
+        let var_name = self.name.map(|v| format!("{v}.size.addr")).unwrap_or_default();
+
+        unsafe {
+            ctx.builder
+                .build_in_bounds_gep(
+                    self.as_base_value(),
+                    &[llvm_i32.const_zero(), llvm_i32.const_int(1, true)],
+                    var_name.as_str(),
+                )
+                .unwrap()
+        }
+    }
+
+    /// Stores the array of data elements `data` into this instance.
+    fn store_data(&self, ctx: &CodeGenContext<'ctx, '_>, data: PointerValue<'ctx>) {
+        ctx.builder.build_store(self.pptr_to_data(ctx), data).unwrap();
+    }
+
+    /// Convenience method for creating a new array storing data elements with the given element
+    /// type `elem_ty` and `size`.
+    ///
+    /// If `size` is [None], the size stored in the field of this instance is used instead.
+    pub fn create_data(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        elem_ty: BasicTypeEnum<'ctx>,
+        size: Option<IntValue<'ctx>>,
+    ) {
+        let size = size.unwrap_or_else(|| self.load_size(ctx, None));
+
+        let data = ctx
+            .builder
+            .build_select(
+                ctx.builder
+                    .build_int_compare(IntPredicate::NE, size, self.llvm_usize.const_zero(), "")
+                    .unwrap(),
+                ctx.builder.build_array_alloca(elem_ty, size, "").unwrap(),
+                elem_ty.ptr_type(AddressSpace::default()).const_zero(),
+                "",
+            )
+            .map(BasicValueEnum::into_pointer_value)
+            .unwrap();
+        self.store_data(ctx, data);
+    }
+
+    /// Returns the double-indirection pointer to the `data` array, as if by calling `getelementptr`
+    /// on the field.
+    #[must_use]
+    pub fn data(&self) -> ListDataProxy<'ctx, '_> {
+        ListDataProxy(self)
+    }
+
+    /// Stores the `size` of this `list` into this instance.
+    pub fn store_size<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+        size: IntValue<'ctx>,
+    ) {
+        debug_assert_eq!(size.get_type(), generator.get_size_type(ctx.ctx));
+
+        let psize = self.ptr_to_size(ctx);
+        ctx.builder.build_store(psize, size).unwrap();
+    }
+
+    /// Returns the size of this `list` as a value.
+    pub fn load_size(&self, ctx: &CodeGenContext<'ctx, '_>, name: Option<&str>) -> IntValue<'ctx> {
+        let psize = self.ptr_to_size(ctx);
+        let var_name = name
+            .map(ToString::to_string)
+            .or_else(|| self.name.map(|v| format!("{v}.size")))
+            .unwrap_or_default();
+
+        ctx.builder
+            .build_load(psize, var_name.as_str())
+            .map(BasicValueEnum::into_int_value)
+            .unwrap()
+    }
+}
+
+impl<'ctx> ProxyValue<'ctx> for ListValue<'ctx> {
+    type Base = PointerValue<'ctx>;
+    type Type = ListType<'ctx>;
+
+    fn get_type(&self) -> Self::Type {
+        ListType::from_type(self.as_base_value().get_type(), self.llvm_usize)
+    }
+
+    fn as_base_value(&self) -> Self::Base {
+        self.value
+    }
+}
+
+impl<'ctx> From<ListValue<'ctx>> for PointerValue<'ctx> {
+    fn from(value: ListValue<'ctx>) -> Self {
+        value.as_base_value()
+    }
+}
+
+/// Proxy type for accessing the `data` array of an `list` instance in LLVM.
+#[derive(Copy, Clone)]
+pub struct ListDataProxy<'ctx, 'a>(&'a ListValue<'ctx>);
+
+impl<'ctx> ArrayLikeValue<'ctx> for ListDataProxy<'ctx, '_> {
+    fn element_type<G: CodeGenerator + ?Sized>(
+        &self,
+        _: &CodeGenContext<'ctx, '_>,
+        _: &G,
+    ) -> AnyTypeEnum<'ctx> {
+        self.0.value.get_type().get_element_type()
+    }
+
+    fn base_ptr<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        _: &G,
+    ) -> PointerValue<'ctx> {
+        let var_name = self.0.name.map(|v| format!("{v}.data")).unwrap_or_default();
+
+        ctx.builder
+            .build_load(self.0.pptr_to_data(ctx), var_name.as_str())
+            .map(BasicValueEnum::into_pointer_value)
+            .unwrap()
+    }
+
+    fn size<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        _: &G,
+    ) -> IntValue<'ctx> {
+        self.0.load_size(ctx, None)
+    }
+}
+
+impl<'ctx> ArrayLikeIndexer<'ctx> for ListDataProxy<'ctx, '_> {
+    unsafe fn ptr_offset_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &IntValue<'ctx>,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        let var_name = name.map(|v| format!("{v}.addr")).unwrap_or_default();
+
+        unsafe {
+            ctx.builder
+                .build_in_bounds_gep(self.base_ptr(ctx, generator), &[*idx], var_name.as_str())
+                .unwrap()
+        }
+    }
+
+    fn ptr_offset<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &IntValue<'ctx>,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        debug_assert_eq!(idx.get_type(), generator.get_size_type(ctx.ctx));
+
+        let size = self.size(ctx, generator);
+        let in_range = ctx.builder.build_int_compare(IntPredicate::ULT, *idx, size, "").unwrap();
+        ctx.make_assert(
+            generator,
+            in_range,
+            "0:IndexError",
+            "list index out of range",
+            [None, None, None],
+            ctx.current_loc,
+        );
+
+        unsafe { self.ptr_offset_unchecked(ctx, generator, idx, name) }
+    }
+}
+
+impl<'ctx> UntypedArrayLikeAccessor<'ctx> for ListDataProxy<'ctx, '_> {}
+impl<'ctx> UntypedArrayLikeMutator<'ctx> for ListDataProxy<'ctx, '_> {}

nac3core/src/codegen/values/mod.rs

@@ -0,0 +1,47 @@
+use inkwell::{context::Context, values::BasicValue};
+
+use super::types::ProxyType;
+use crate::codegen::CodeGenerator;
+pub use array::*;
+pub use list::*;
+pub use ndarray::*;
+pub use range::*;
+
+mod array;
+mod list;
+mod ndarray;
+mod range;
+
+/// A LLVM type that is used to represent a non-primitive value in NAC3.
+pub trait ProxyValue<'ctx>: Into<Self::Base> {
+    /// The type of LLVM values represented by this instance. This is usually the
+    /// [LLVM pointer type][PointerValue].
+    type Base: BasicValue<'ctx>;
+
+    /// The type of this value.
+    type Type: ProxyType<'ctx, Value = Self>;
+
+    /// Checks whether `value` can be represented by this [`ProxyValue`].
+    fn is_instance<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        value: impl BasicValue<'ctx>,
+    ) -> Result<(), String> {
+        Self::Type::is_type(generator, ctx, value.as_basic_value_enum().get_type())
+    }
+
+    /// Checks whether `value` can be represented by this [`ProxyValue`].
+    fn is_representable<G: CodeGenerator + ?Sized>(
+        generator: &G,
+        ctx: &'ctx Context,
+        value: Self::Base,
+    ) -> Result<(), String> {
+        Self::is_instance(generator, ctx, value.as_basic_value_enum())
+    }
+
+    /// Returns the [type][ProxyType] of this value.
+    fn get_type(&self) -> Self::Type;
+
+    /// Returns the [base value][Self::Base] of this proxy.
+    fn as_base_value(&self) -> Self::Base;
+}

nac3core/src/codegen/values/ndarray.rs

@@ -0,0 +1,523 @@
+use inkwell::{
+    types::{AnyType, AnyTypeEnum, BasicType, BasicTypeEnum, IntType},
+    values::{BasicValueEnum, IntValue, PointerValue},
+    AddressSpace, IntPredicate,
+};
+
+use super::{
+    ArrayLikeIndexer, ArrayLikeValue, ProxyValue, TypedArrayLikeAccessor, TypedArrayLikeMutator,
+    UntypedArrayLikeAccessor, UntypedArrayLikeMutator,
+};
+use crate::codegen::{
+    irrt::{call_ndarray_calc_size, call_ndarray_flatten_index},
+    llvm_intrinsics::call_int_umin,
+    stmt::gen_for_callback_incrementing,
+    types::NDArrayType,
+    CodeGenContext, CodeGenerator,
+};
+
+/// Proxy type for accessing an `NDArray` value in LLVM.
+#[derive(Copy, Clone)]
+pub struct NDArrayValue<'ctx> {
+    value: PointerValue<'ctx>,
+    dtype: BasicTypeEnum<'ctx>,
+    llvm_usize: IntType<'ctx>,
+    name: Option<&'ctx str>,
+}
+
+impl<'ctx> NDArrayValue<'ctx> {
+    /// Checks whether `value` is an instance of `NDArray`, returning [Err] if `value` is not an
+    /// instance.
+    pub fn is_representable(
+        value: PointerValue<'ctx>,
+        llvm_usize: IntType<'ctx>,
+    ) -> Result<(), String> {
+        NDArrayType::is_representable(value.get_type(), llvm_usize)
+    }
+
+    /// Creates an [`NDArrayValue`] from a [`PointerValue`].
+    #[must_use]
+    pub fn from_pointer_value(
+        ptr: PointerValue<'ctx>,
+        dtype: BasicTypeEnum<'ctx>,
+        llvm_usize: IntType<'ctx>,
+        name: Option<&'ctx str>,
+    ) -> Self {
+        debug_assert!(Self::is_representable(ptr, llvm_usize).is_ok());
+
+        NDArrayValue { value: ptr, dtype, llvm_usize, name }
+    }
+
+    /// Returns the pointer to the field storing the number of dimensions of this `NDArray`.
+    fn ptr_to_ndims(&self, ctx: &CodeGenContext<'ctx, '_>) -> PointerValue<'ctx> {
+        self.get_type()
+            .get_fields(ctx.ctx, self.llvm_usize)
+            .ndims
+            .ptr_by_gep(ctx, self.value, self.name)
+    }
+
+    /// Stores the number of dimensions `ndims` into this instance.
+    pub fn store_ndims<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+        ndims: IntValue<'ctx>,
+    ) {
+        debug_assert_eq!(ndims.get_type(), generator.get_size_type(ctx.ctx));
+
+        let pndims = self.ptr_to_ndims(ctx);
+        ctx.builder.build_store(pndims, ndims).unwrap();
+    }
+
+    /// Returns the number of dimensions of this `NDArray` as a value.
+    pub fn load_ndims(&self, ctx: &CodeGenContext<'ctx, '_>) -> IntValue<'ctx> {
+        let pndims = self.ptr_to_ndims(ctx);
+        ctx.builder.build_load(pndims, "").map(BasicValueEnum::into_int_value).unwrap()
+    }
+
+    /// Returns the double-indirection pointer to the `shape` array, as if by calling
+    /// `getelementptr` on the field.
+    fn ptr_to_shape(&self, ctx: &CodeGenContext<'ctx, '_>) -> PointerValue<'ctx> {
+        self.get_type()
+            .get_fields(ctx.ctx, self.llvm_usize)
+            .shape
+            .ptr_by_gep(ctx, self.value, self.name)
+    }
+
+    /// Stores the array of dimension sizes `dims` into this instance.
+    fn store_shape(&self, ctx: &CodeGenContext<'ctx, '_>, dims: PointerValue<'ctx>) {
+        ctx.builder.build_store(self.ptr_to_shape(ctx), dims).unwrap();
+    }
+
+    /// Convenience method for creating a new array storing dimension sizes with the given `size`.
+    pub fn create_shape(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        llvm_usize: IntType<'ctx>,
+        size: IntValue<'ctx>,
+    ) {
+        self.store_shape(ctx, ctx.builder.build_array_alloca(llvm_usize, size, "").unwrap());
+    }
+
+    /// Returns a proxy object to the field storing the size of each dimension of this `NDArray`.
+    #[must_use]
+    pub fn shape(&self) -> NDArrayShapeProxy<'ctx, '_> {
+        NDArrayShapeProxy(self)
+    }
+
+    /// Returns the double-indirection pointer to the `data` array, as if by calling `getelementptr`
+    /// on the field.
+    pub fn ptr_to_data(&self, ctx: &CodeGenContext<'ctx, '_>) -> PointerValue<'ctx> {
+        self.get_type()
+            .get_fields(ctx.ctx, self.llvm_usize)
+            .data
+            .ptr_by_gep(ctx, self.value, self.name)
+    }
+
+    /// Stores the array of data elements `data` into this instance.
+    fn store_data(&self, ctx: &CodeGenContext<'ctx, '_>, data: PointerValue<'ctx>) {
+        let data = ctx
+            .builder
+            .build_bit_cast(data, ctx.ctx.i8_type().ptr_type(AddressSpace::default()), "")
+            .unwrap();
+        ctx.builder.build_store(self.ptr_to_data(ctx), data).unwrap();
+    }
+
+    /// Convenience method for creating a new array storing data elements with the given element
+    /// type `elem_ty` and `size`.
+    pub fn create_data(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        elem_ty: BasicTypeEnum<'ctx>,
+        size: IntValue<'ctx>,
+    ) {
+        let itemsize =
+            ctx.builder.build_int_cast(elem_ty.size_of().unwrap(), size.get_type(), "").unwrap();
+        let nbytes = ctx.builder.build_int_mul(size, itemsize, "").unwrap();
+
+        // TODO: What about alignment?
+        self.store_data(
+            ctx,
+            ctx.builder.build_array_alloca(ctx.ctx.i8_type(), nbytes, "").unwrap(),
+        );
+    }
+
+    /// Returns a proxy object to the field storing the data of this `NDArray`.
+    #[must_use]
+    pub fn data(&self) -> NDArrayDataProxy<'ctx, '_> {
+        NDArrayDataProxy(self)
+    }
+}
+
+impl<'ctx> ProxyValue<'ctx> for NDArrayValue<'ctx> {
+    type Base = PointerValue<'ctx>;
+    type Type = NDArrayType<'ctx>;
+
+    fn get_type(&self) -> Self::Type {
+        NDArrayType::from_type(self.as_base_value().get_type(), self.dtype, self.llvm_usize)
+    }
+
+    fn as_base_value(&self) -> Self::Base {
+        self.value
+    }
+}
+
+impl<'ctx> From<NDArrayValue<'ctx>> for PointerValue<'ctx> {
+    fn from(value: NDArrayValue<'ctx>) -> Self {
+        value.as_base_value()
+    }
+}
+
+/// Proxy type for accessing the `dims` array of an `NDArray` instance in LLVM.
+#[derive(Copy, Clone)]
+pub struct NDArrayShapeProxy<'ctx, 'a>(&'a NDArrayValue<'ctx>);
+
+impl<'ctx> ArrayLikeValue<'ctx> for NDArrayShapeProxy<'ctx, '_> {
+    fn element_type<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+    ) -> AnyTypeEnum<'ctx> {
+        self.0.shape().base_ptr(ctx, generator).get_type().get_element_type()
+    }
+
+    fn base_ptr<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        _: &G,
+    ) -> PointerValue<'ctx> {
+        let var_name = self.0.name.map(|v| format!("{v}.data")).unwrap_or_default();
+
+        ctx.builder
+            .build_load(self.0.ptr_to_shape(ctx), var_name.as_str())
+            .map(BasicValueEnum::into_pointer_value)
+            .unwrap()
+    }
+
+    fn size<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        _: &G,
+    ) -> IntValue<'ctx> {
+        self.0.load_ndims(ctx)
+    }
+}
+
+impl<'ctx> ArrayLikeIndexer<'ctx, IntValue<'ctx>> for NDArrayShapeProxy<'ctx, '_> {
+    unsafe fn ptr_offset_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &IntValue<'ctx>,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        let var_name = name.map(|v| format!("{v}.addr")).unwrap_or_default();
+
+        unsafe {
+            ctx.builder
+                .build_in_bounds_gep(self.base_ptr(ctx, generator), &[*idx], var_name.as_str())
+                .unwrap()
+        }
+    }
+
+    fn ptr_offset<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &IntValue<'ctx>,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        let size = self.size(ctx, generator);
+        let in_range = ctx.builder.build_int_compare(IntPredicate::ULT, *idx, size, "").unwrap();
+        ctx.make_assert(
+            generator,
+            in_range,
+            "0:IndexError",
+            "index {0} is out of bounds for axis 0 with size {1}",
+            [Some(*idx), Some(self.0.load_ndims(ctx)), None],
+            ctx.current_loc,
+        );
+
+        unsafe { self.ptr_offset_unchecked(ctx, generator, idx, name) }
+    }
+}
+
+impl<'ctx> UntypedArrayLikeAccessor<'ctx, IntValue<'ctx>> for NDArrayShapeProxy<'ctx, '_> {}
+impl<'ctx> UntypedArrayLikeMutator<'ctx, IntValue<'ctx>> for NDArrayShapeProxy<'ctx, '_> {}
+
+impl<'ctx> TypedArrayLikeAccessor<'ctx, IntValue<'ctx>> for NDArrayShapeProxy<'ctx, '_> {
+    fn downcast_to_type(
+        &self,
+        _: &mut CodeGenContext<'ctx, '_>,
+        value: BasicValueEnum<'ctx>,
+    ) -> IntValue<'ctx> {
+        value.into_int_value()
+    }
+}
+
+impl<'ctx> TypedArrayLikeMutator<'ctx, IntValue<'ctx>> for NDArrayShapeProxy<'ctx, '_> {
+    fn upcast_from_type(
+        &self,
+        _: &mut CodeGenContext<'ctx, '_>,
+        value: IntValue<'ctx>,
+    ) -> BasicValueEnum<'ctx> {
+        value.into()
+    }
+}
+
+/// Proxy type for accessing the `data` array of an `NDArray` instance in LLVM.
+#[derive(Copy, Clone)]
+pub struct NDArrayDataProxy<'ctx, 'a>(&'a NDArrayValue<'ctx>);
+
+impl<'ctx> ArrayLikeValue<'ctx> for NDArrayDataProxy<'ctx, '_> {
+    fn element_type<G: CodeGenerator + ?Sized>(
+        &self,
+        _: &CodeGenContext<'ctx, '_>,
+        _: &G,
+    ) -> AnyTypeEnum<'ctx> {
+        self.0.dtype.as_any_type_enum()
+    }
+
+    fn base_ptr<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        _: &G,
+    ) -> PointerValue<'ctx> {
+        let var_name = self.0.name.map(|v| format!("{v}.data")).unwrap_or_default();
+
+        ctx.builder
+            .build_load(self.0.ptr_to_data(ctx), var_name.as_str())
+            .map(BasicValueEnum::into_pointer_value)
+            .unwrap()
+    }
+
+    fn size<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        generator: &G,
+    ) -> IntValue<'ctx> {
+        call_ndarray_calc_size(generator, ctx, &self.as_slice_value(ctx, generator), (None, None))
+    }
+}
+
+impl<'ctx> ArrayLikeIndexer<'ctx> for NDArrayDataProxy<'ctx, '_> {
+    unsafe fn ptr_offset_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &IntValue<'ctx>,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        let sizeof_elem = ctx
+            .builder
+            .build_int_truncate_or_bit_cast(
+                self.element_type(ctx, generator).size_of().unwrap(),
+                idx.get_type(),
+                "",
+            )
+            .unwrap();
+        let idx = ctx.builder.build_int_mul(*idx, sizeof_elem, "").unwrap();
+        let ptr = unsafe {
+            ctx.builder
+                .build_in_bounds_gep(
+                    self.base_ptr(ctx, generator),
+                    &[idx],
+                    name.unwrap_or_default(),
+                )
+                .unwrap()
+        };
+
+        // Current implementation is transparent - The returned pointer type is
+        // already cast into the expected type, allowing for immediately
+        // load/store.
+        ctx.builder
+            .build_pointer_cast(
+                ptr,
+                BasicTypeEnum::try_from(self.element_type(ctx, generator))
+                    .unwrap()
+                    .ptr_type(AddressSpace::default()),
+                "",
+            )
+            .unwrap()
+    }
+
+    fn ptr_offset<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        idx: &IntValue<'ctx>,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        let data_sz = self.size(ctx, generator);
+        let in_range = ctx.builder.build_int_compare(IntPredicate::ULT, *idx, data_sz, "").unwrap();
+        ctx.make_assert(
+            generator,
+            in_range,
+            "0:IndexError",
+            "index {0} is out of bounds with size {1}",
+            [Some(*idx), Some(self.0.load_ndims(ctx)), None],
+            ctx.current_loc,
+        );
+
+        let ptr = unsafe { self.ptr_offset_unchecked(ctx, generator, idx, name) };
+
+        // Current implementation is transparent - The returned pointer type is
+        // already cast into the expected type, allowing for immediately
+        // load/store.
+        ctx.builder
+            .build_pointer_cast(
+                ptr,
+                BasicTypeEnum::try_from(self.element_type(ctx, generator))
+                    .unwrap()
+                    .ptr_type(AddressSpace::default()),
+                "",
+            )
+            .unwrap()
+    }
+}
+
+impl<'ctx> UntypedArrayLikeAccessor<'ctx, IntValue<'ctx>> for NDArrayDataProxy<'ctx, '_> {}
+impl<'ctx> UntypedArrayLikeMutator<'ctx, IntValue<'ctx>> for NDArrayDataProxy<'ctx, '_> {}
+
+impl<'ctx, Index: UntypedArrayLikeAccessor<'ctx>> ArrayLikeIndexer<'ctx, Index>
+    for NDArrayDataProxy<'ctx, '_>
+{
+    unsafe fn ptr_offset_unchecked<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        indices: &Index,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        let llvm_usize = generator.get_size_type(ctx.ctx);
+
+        let indices_elem_ty = indices
+            .ptr_offset(ctx, generator, &llvm_usize.const_zero(), None)
+            .get_type()
+            .get_element_type();
+        let Ok(indices_elem_ty) = IntType::try_from(indices_elem_ty) else {
+            panic!("Expected list[int32] but got {indices_elem_ty}")
+        };
+        assert_eq!(
+            indices_elem_ty.get_bit_width(),
+            32,
+            "Expected list[int32] but got list[int{}]",
+            indices_elem_ty.get_bit_width()
+        );
+
+        let index = call_ndarray_flatten_index(generator, ctx, *self.0, indices);
+        let sizeof_elem = ctx
+            .builder
+            .build_int_truncate_or_bit_cast(
+                self.element_type(ctx, generator).size_of().unwrap(),
+                index.get_type(),
+                "",
+            )
+            .unwrap();
+        let index = ctx.builder.build_int_mul(index, sizeof_elem, "").unwrap();
+
+        let ptr = unsafe {
+            ctx.builder
+                .build_in_bounds_gep(
+                    self.base_ptr(ctx, generator),
+                    &[index],
+                    name.unwrap_or_default(),
+                )
+                .unwrap()
+        };
+        // TODO: Current implementation is transparent
+        ctx.builder
+            .build_pointer_cast(
+                ptr,
+                BasicTypeEnum::try_from(self.element_type(ctx, generator))
+                    .unwrap()
+                    .ptr_type(AddressSpace::default()),
+                "",
+            )
+            .unwrap()
+    }
+
+    fn ptr_offset<G: CodeGenerator + ?Sized>(
+        &self,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut G,
+        indices: &Index,
+        name: Option<&str>,
+    ) -> PointerValue<'ctx> {
+        let llvm_usize = generator.get_size_type(ctx.ctx);
+
+        let indices_size = indices.size(ctx, generator);
+        let nidx_leq_ndims = ctx
+            .builder
+            .build_int_compare(IntPredicate::SLE, indices_size, self.0.load_ndims(ctx), "")
+            .unwrap();
+        ctx.make_assert(
+            generator,
+            nidx_leq_ndims,
+            "0:IndexError",
+            "invalid index to scalar variable",
+            [None, None, None],
+            ctx.current_loc,
+        );
+
+        let indices_len = indices.size(ctx, generator);
+        let ndarray_len = self.0.load_ndims(ctx);
+        let len = call_int_umin(ctx, indices_len, ndarray_len, None);
+        gen_for_callback_incrementing(
+            generator,
+            ctx,
+            None,
+            llvm_usize.const_zero(),
+            (len, false),
+            |generator, ctx, _, i| {
+                let (dim_idx, dim_sz) = unsafe {
+                    (
+                        indices.get_unchecked(ctx, generator, &i, None).into_int_value(),
+                        self.0.shape().get_typed_unchecked(ctx, generator, &i, None),
+                    )
+                };
+                let dim_idx = ctx
+                    .builder
+                    .build_int_z_extend_or_bit_cast(dim_idx, dim_sz.get_type(), "")
+                    .unwrap();
+
+                let dim_lt =
+                    ctx.builder.build_int_compare(IntPredicate::SLT, dim_idx, dim_sz, "").unwrap();
+
+                ctx.make_assert(
+                    generator,
+                    dim_lt,
+                    "0:IndexError",
+                    "index {0} is out of bounds for axis 0 with size {1}",
+                    [Some(dim_idx), Some(dim_sz), None],
+                    ctx.current_loc,
+                );
+
+                Ok(())
+            },
+            llvm_usize.const_int(1, false),
+        )
+        .unwrap();
+
+        let ptr = unsafe { self.ptr_offset_unchecked(ctx, generator, indices, name) };
+        // TODO: Current implementation is transparent
+        ctx.builder
+            .build_pointer_cast(
+                ptr,
+                BasicTypeEnum::try_from(self.element_type(ctx, generator))
+                    .unwrap()
+                    .ptr_type(AddressSpace::default()),
+                "",
+            )
+            .unwrap()
+    }
+}
+
+impl<'ctx, Index: UntypedArrayLikeAccessor<'ctx>> UntypedArrayLikeAccessor<'ctx, Index>
+    for NDArrayDataProxy<'ctx, '_>
+{
+}
+impl<'ctx, Index: UntypedArrayLikeAccessor<'ctx>> UntypedArrayLikeMutator<'ctx, Index>
+    for NDArrayDataProxy<'ctx, '_>
+{
+}

nac3core/src/codegen/values/range.rs

@@ -0,0 +1,153 @@
+use inkwell::values::{BasicValueEnum, IntValue, PointerValue};
+
+use super::ProxyValue;
+use crate::codegen::{types::RangeType, CodeGenContext};
+
+/// Proxy type for accessing a `range` value in LLVM.
+#[derive(Copy, Clone)]
+pub struct RangeValue<'ctx> {
+    value: PointerValue<'ctx>,
+    name: Option<&'ctx str>,
+}
+
+impl<'ctx> RangeValue<'ctx> {
+    /// Checks whether `value` is an instance of `range`, returning [Err] if `value` is not an instance.
+    pub fn is_representable(value: PointerValue<'ctx>) -> Result<(), String> {
+        RangeType::is_representable(value.get_type())
+    }
+
+    /// Creates an [`RangeValue`] from a [`PointerValue`].
+    #[must_use]
+    pub fn from_pointer_value(ptr: PointerValue<'ctx>, name: Option<&'ctx str>) -> Self {
+        debug_assert!(Self::is_representable(ptr).is_ok());
+
+        RangeValue { value: ptr, name }
+    }
+
+    fn ptr_to_start(&self, ctx: &CodeGenContext<'ctx, '_>) -> PointerValue<'ctx> {
+        let llvm_i32 = ctx.ctx.i32_type();
+        let var_name = self.name.map(|v| format!("{v}.start.addr")).unwrap_or_default();
+
+        unsafe {
+            ctx.builder
+                .build_in_bounds_gep(
+                    self.as_base_value(),
+                    &[llvm_i32.const_zero(), llvm_i32.const_int(0, false)],
+                    var_name.as_str(),
+                )
+                .unwrap()
+        }
+    }
+
+    fn ptr_to_end(&self, ctx: &CodeGenContext<'ctx, '_>) -> PointerValue<'ctx> {
+        let llvm_i32 = ctx.ctx.i32_type();
+        let var_name = self.name.map(|v| format!("{v}.end.addr")).unwrap_or_default();
+
+        unsafe {
+            ctx.builder
+                .build_in_bounds_gep(
+                    self.as_base_value(),
+                    &[llvm_i32.const_zero(), llvm_i32.const_int(1, false)],
+                    var_name.as_str(),
+                )
+                .unwrap()
+        }
+    }
+
+    fn ptr_to_step(&self, ctx: &CodeGenContext<'ctx, '_>) -> PointerValue<'ctx> {
+        let llvm_i32 = ctx.ctx.i32_type();
+        let var_name = self.name.map(|v| format!("{v}.step.addr")).unwrap_or_default();
+
+        unsafe {
+            ctx.builder
+                .build_in_bounds_gep(
+                    self.as_base_value(),
+                    &[llvm_i32.const_zero(), llvm_i32.const_int(2, false)],
+                    var_name.as_str(),
+                )
+                .unwrap()
+        }
+    }
+
+    /// Stores the `start` value into this instance.
+    pub fn store_start(&self, ctx: &CodeGenContext<'ctx, '_>, start: IntValue<'ctx>) {
+        debug_assert_eq!(start.get_type().get_bit_width(), 32);
+
+        let pstart = self.ptr_to_start(ctx);
+        ctx.builder.build_store(pstart, start).unwrap();
+    }
+
+    /// Returns the `start` value of this `range`.
+    pub fn load_start(&self, ctx: &CodeGenContext<'ctx, '_>, name: Option<&str>) -> IntValue<'ctx> {
+        let pstart = self.ptr_to_start(ctx);
+        let var_name = name
+            .map(ToString::to_string)
+            .or_else(|| self.name.map(|v| format!("{v}.start")))
+            .unwrap_or_default();
+
+        ctx.builder
+            .build_load(pstart, var_name.as_str())
+            .map(BasicValueEnum::into_int_value)
+            .unwrap()
+    }
+
+    /// Stores the `end` value into this instance.
+    pub fn store_end(&self, ctx: &CodeGenContext<'ctx, '_>, end: IntValue<'ctx>) {
+        debug_assert_eq!(end.get_type().get_bit_width(), 32);
+
+        let pend = self.ptr_to_end(ctx);
+        ctx.builder.build_store(pend, end).unwrap();
+    }
+
+    /// Returns the `end` value of this `range`.
+    pub fn load_end(&self, ctx: &CodeGenContext<'ctx, '_>, name: Option<&str>) -> IntValue<'ctx> {
+        let pend = self.ptr_to_end(ctx);
+        let var_name = name
+            .map(ToString::to_string)
+            .or_else(|| self.name.map(|v| format!("{v}.end")))
+            .unwrap_or_default();
+
+        ctx.builder.build_load(pend, var_name.as_str()).map(BasicValueEnum::into_int_value).unwrap()
+    }
+
+    /// Stores the `step` value into this instance.
+    pub fn store_step(&self, ctx: &CodeGenContext<'ctx, '_>, step: IntValue<'ctx>) {
+        debug_assert_eq!(step.get_type().get_bit_width(), 32);
+
+        let pstep = self.ptr_to_step(ctx);
+        ctx.builder.build_store(pstep, step).unwrap();
+    }
+
+    /// Returns the `step` value of this `range`.
+    pub fn load_step(&self, ctx: &CodeGenContext<'ctx, '_>, name: Option<&str>) -> IntValue<'ctx> {
+        let pstep = self.ptr_to_step(ctx);
+        let var_name = name
+            .map(ToString::to_string)
+            .or_else(|| self.name.map(|v| format!("{v}.step")))
+            .unwrap_or_default();
+
+        ctx.builder
+            .build_load(pstep, var_name.as_str())
+            .map(BasicValueEnum::into_int_value)
+            .unwrap()
+    }
+}
+
+impl<'ctx> ProxyValue<'ctx> for RangeValue<'ctx> {
+    type Base = PointerValue<'ctx>;
+    type Type = RangeType<'ctx>;
+
+    fn get_type(&self) -> Self::Type {
+        RangeType::from_type(self.value.get_type())
+    }
+
+    fn as_base_value(&self) -> Self::Base {
+        self.value
+    }
+}
+
+impl<'ctx> From<RangeValue<'ctx>> for PointerValue<'ctx> {
+    fn from(value: RangeValue<'ctx>) -> Self {
+        value.as_base_value()
+    }
+}

nac3core/src/lib.rs

@@ -1,10 +1,4 @@
-#![deny(
-    future_incompatible,
-    let_underscore,
-    nonstandard_style,
-    rust_2024_compatibility,
-    clippy::all
-)]
+#![deny(future_incompatible, let_underscore, nonstandard_style, clippy::all)]
 #![warn(clippy::pedantic)]
 #![allow(
     dead_code,
@@ -19,8 +13,13 @@
     clippy::wildcard_imports
 )]
 
+// users of nac3core need to use the same version of these dependencies, so expose them as nac3core::*
+pub use inkwell;
+pub use nac3parser;
+
 pub mod codegen;
 pub mod symbol_resolver;
 pub mod toplevel;
 pub mod typecheck;
-pub mod util;
+
+extern crate self as nac3core;

nac3core/src/symbol_resolver.rs

@@ -1,7 +1,15 @@
-use std::fmt::Debug;
-use std::rc::Rc;
-use std::sync::Arc;
-use std::{collections::HashMap, collections::HashSet, fmt::Display};
+use std::{
+    collections::{HashMap, HashSet},
+    fmt::{Debug, Display},
+    rc::Rc,
+    sync::Arc,
+};
+
+use inkwell::values::{BasicValueEnum, FloatValue, IntValue, PointerValue, StructValue};
+use itertools::{chain, izip, Itertools};
+use parking_lot::RwLock;
+
+use nac3parser::ast::{Constant, Expr, Location, StrRef};
 
 use crate::{
     codegen::{CodeGenContext, CodeGenerator},
@@ -11,10 +19,6 @@ use crate::{
         typedef::{Type, TypeEnum, Unifier, VarMap},
     },
 };
-use inkwell::values::{BasicValueEnum, FloatValue, IntValue, PointerValue, StructValue};
-use itertools::{chain, izip, Itertools};
-use nac3parser::ast::{Constant, Expr, Location, StrRef};
-use parking_lot::RwLock;
 
 #[derive(Clone, PartialEq, Debug)]
 pub enum SymbolValue {
@@ -78,14 +82,14 @@ impl SymbolValue {
             }
             Constant::Tuple(t) => {
                 let expected_ty = unifier.get_ty(expected_ty);
-                let TypeEnum::TTuple { ty } = expected_ty.as_ref() else {
+                let TypeEnum::TTuple { ty, is_vararg_ctx } = expected_ty.as_ref() else {
                     return Err(format!(
                         "Expected {:?}, but got Tuple",
                         expected_ty.get_type_name()
                     ));
                 };
 
-                assert_eq!(ty.len(), t.len());
+                assert!(*is_vararg_ctx || ty.len() == t.len());
 
                 let elems = t
                     .iter()
@@ -155,7 +159,7 @@ impl SymbolValue {
             SymbolValue::Bool(_) => primitives.bool,
             SymbolValue::Tuple(vs) => {
                 let vs_tys = vs.iter().map(|v| v.get_type(primitives, unifier)).collect::<Vec<_>>();
-                unifier.add_ty(TypeEnum::TTuple { ty: vs_tys })
+                unifier.add_ty(TypeEnum::TTuple { ty: vs_tys, is_vararg_ctx: false })
             }
             SymbolValue::OptionSome(_) | SymbolValue::OptionNone => primitives.option,
         }
@@ -365,6 +369,7 @@ pub trait SymbolResolver {
         &self,
         str: StrRef,
         ctx: &mut CodeGenContext<'ctx, '_>,
+        generator: &mut dyn CodeGenerator,
     ) -> Option<ValueEnum<'ctx>>;
 
     fn get_default_param_value(&self, expr: &Expr) -> Option<SymbolValue>;
@@ -482,7 +487,7 @@ pub fn parse_type_annotation<T>(
                         parse_type_annotation(resolver, top_level_defs, unifier, primitives, elt)
                     })
                     .collect::<Result<Vec<_>, _>>()?;
-                Ok(unifier.add_ty(TypeEnum::TTuple { ty }))
+                Ok(unifier.add_ty(TypeEnum::TTuple { ty, is_vararg_ctx: false }))
             } else {
                 Err(HashSet::from(["Expected multiple elements for tuple".into()]))
             }

nac3core/src/toplevel/builtins.rs

@@ -1,7 +1,5 @@
 use std::iter::once;
 
-use crate::util::SizeVariant;
-use helper::{debug_assert_prim_is_allowed, make_exception_fields, PrimDefDetails};
 use indexmap::IndexMap;
 use inkwell::{
     attributes::{Attribute, AttributeLoc},
@@ -12,22 +10,22 @@ use inkwell::{
 use itertools::Either;
 use strum::IntoEnumIterator;
 
+use super::{
+    helper::{debug_assert_prim_is_allowed, make_exception_fields, PrimDef, PrimDefDetails},
+    numpy::make_ndarray_ty,
+    *,
+};
 use crate::{
     codegen::{
         builtin_fns,
-        classes::{ArrayLikeValue, NDArrayValue, ProxyValue, RangeValue, TypedArrayLikeAccessor},
-        expr::destructure_range,
-        irrt::*,
         numpy::*,
         stmt::exn_constructor,
+        values::{ProxyValue, RangeValue},
     },
     symbol_resolver::SymbolValue,
-    toplevel::{helper::PrimDef, numpy::make_ndarray_ty},
     typecheck::typedef::{into_var_map, iter_type_vars, TypeVar, VarMap},
 };
 
-use super::*;
-
 type BuiltinInfo = Vec<(Arc<RwLock<TopLevelDef>>, Option<Stmt>)>;
 
 pub fn get_exn_constructor(
@@ -46,10 +44,26 @@ pub fn get_exn_constructor(
             name: "msg".into(),
             ty: string,
             default_value: Some(SymbolValue::Str(String::new())),
+            is_vararg: false,
+        },
+        FuncArg {
+            name: "param0".into(),
+            ty: int64,
+            default_value: Some(SymbolValue::I64(0)),
+            is_vararg: false,
+        },
+        FuncArg {
+            name: "param1".into(),
+            ty: int64,
+            default_value: Some(SymbolValue::I64(0)),
+            is_vararg: false,
+        },
+        FuncArg {
+            name: "param2".into(),
+            ty: int64,
+            default_value: Some(SymbolValue::I64(0)),
+            is_vararg: false,
         },
-        FuncArg { name: "param0".into(), ty: int64, default_value: Some(SymbolValue::I64(0)) },
-        FuncArg { name: "param1".into(), ty: int64, default_value: Some(SymbolValue::I64(0)) },
-        FuncArg { name: "param2".into(), ty: int64, default_value: Some(SymbolValue::I64(0)) },
     ];
     let exn_type = unifier.add_ty(TypeEnum::TObj {
         obj_id: DefinitionId(class_id),
@@ -115,7 +129,12 @@ fn create_fn_by_codegen(
         signature: unifier.add_ty(TypeEnum::TFunc(FunSignature {
             args: param_ty
                 .iter()
-                .map(|p| FuncArg { name: p.1.into(), ty: p.0, default_value: None })
+                .map(|p| FuncArg {
+                    name: p.1.into(),
+                    ty: p.0,
+                    default_value: None,
+                    is_vararg: false,
+                })
                 .collect(),
             ret: ret_ty,
             vars: var_map.clone(),
@@ -279,12 +298,21 @@ pub fn get_builtins(unifier: &mut Unifier, primitives: &PrimitiveStore) -> Built
         .collect()
 }
 
-fn size_variant_to_int_type(variant: SizeVariant, primitives: &PrimitiveStore) -> Type {
-    match variant {
+/// A helper enum used by [`BuiltinBuilder`]
+#[derive(Clone, Copy)]
+enum SizeVariant {
+    Bits32,
+    Bits64,
+}
+
+impl SizeVariant {
+    fn of_int(self, primitives: &PrimitiveStore) -> Type {
+        match self {
             SizeVariant::Bits32 => primitives.int32,
             SizeVariant::Bits64 => primitives.int64,
         }
     }
+}
 
 struct BuiltinBuilder<'a> {
     unifier: &'a mut Unifier,
@@ -338,8 +366,8 @@ impl<'a> BuiltinBuilder<'a> {
         let (is_some_ty, unwrap_ty, option_tvar) =
             if let TypeEnum::TObj { fields, params, .. } = unifier.get_ty(option).as_ref() {
                 (
-                    *fields.get(&PrimDef::OptionIsSome.simple_name().into()).unwrap(),
-                    *fields.get(&PrimDef::OptionUnwrap.simple_name().into()).unwrap(),
+                    *fields.get(&PrimDef::FunOptionIsSome.simple_name().into()).unwrap(),
+                    *fields.get(&PrimDef::FunOptionUnwrap.simple_name().into()).unwrap(),
                     iter_type_vars(params).next().unwrap(),
                 )
             } else {
@@ -354,9 +382,9 @@ impl<'a> BuiltinBuilder<'a> {
         let ndarray_dtype_tvar = iter_type_vars(ndarray_params).next().unwrap();
         let ndarray_ndims_tvar = iter_type_vars(ndarray_params).nth(1).unwrap();
         let ndarray_copy_ty =
-            *ndarray_fields.get(&PrimDef::NDArrayCopy.simple_name().into()).unwrap();
+            *ndarray_fields.get(&PrimDef::FunNDArrayCopy.simple_name().into()).unwrap();
         let ndarray_fill_ty =
-            *ndarray_fields.get(&PrimDef::NDArrayFill.simple_name().into()).unwrap();
+            *ndarray_fields.get(&PrimDef::FunNDArrayFill.simple_name().into()).unwrap();
 
         let num_ty = unifier.get_fresh_var_with_range(
             &[int32, int64, float, boolean, uint32, uint64],
@@ -456,14 +484,14 @@ impl<'a> BuiltinBuilder<'a> {
             PrimDef::Exception => self.build_exception_class_related(prim),
 
             PrimDef::Option
-            | PrimDef::OptionIsSome
-            | PrimDef::OptionIsNone
-            | PrimDef::OptionUnwrap
+            | PrimDef::FunOptionIsSome
+            | PrimDef::FunOptionIsNone
+            | PrimDef::FunOptionUnwrap
             | PrimDef::FunSome => self.build_option_class_related(prim),
 
             PrimDef::List => self.build_list_class_related(prim),
 
-            PrimDef::NDArray | PrimDef::NDArrayCopy | PrimDef::NDArrayFill => {
+            PrimDef::NDArray | PrimDef::FunNDArrayCopy | PrimDef::FunNDArrayFill => {
                 self.build_ndarray_class_related(prim)
             }
 
@@ -502,7 +530,9 @@ impl<'a> BuiltinBuilder<'a> {
 
             PrimDef::FunMin | PrimDef::FunMax => self.build_min_max_function(prim),
 
-            PrimDef::FunNpMin | PrimDef::FunNpMax => self.build_np_min_max_function(prim),
+            PrimDef::FunNpArgmin | PrimDef::FunNpArgmax | PrimDef::FunNpMin | PrimDef::FunNpMax => {
+                self.build_np_max_min_function(prim)
+            }
 
             PrimDef::FunNpMinimum | PrimDef::FunNpMaximum => {
                 self.build_np_minimum_maximum_function(prim)
@@ -546,6 +576,22 @@ impl<'a> BuiltinBuilder<'a> {
             | PrimDef::FunNpLdExp
             | PrimDef::FunNpHypot
             | PrimDef::FunNpNextAfter => self.build_np_2ary_function(prim),
+
+            PrimDef::FunNpTranspose | PrimDef::FunNpReshape => {
+                self.build_np_sp_ndarray_function(prim)
+            }
+
+            PrimDef::FunNpDot
+            | PrimDef::FunNpLinalgCholesky
+            | PrimDef::FunNpLinalgQr
+            | PrimDef::FunNpLinalgSvd
+            | PrimDef::FunNpLinalgInv
+            | PrimDef::FunNpLinalgPinv
+            | PrimDef::FunNpLinalgMatrixPower
+            | PrimDef::FunNpLinalgDet
+            | PrimDef::FunSpLinalgLu
+            | PrimDef::FunSpLinalgSchur
+            | PrimDef::FunSpLinalgHessenberg => self.build_linalg_methods(prim),
         };
 
         if cfg!(debug_assertions) {
@@ -554,7 +600,7 @@ impl<'a> BuiltinBuilder<'a> {
             match (&tld, prim.details()) {
                 (
                     TopLevelDef::Class { name, object_id, .. },
-                    PrimDefDetails::PrimClass { name: exp_name },
+                    PrimDefDetails::PrimClass { name: exp_name, .. },
                 ) => {
                     let exp_object_id = prim.id();
                     assert_eq!(name, &exp_name.into());
@@ -603,17 +649,24 @@ impl<'a> BuiltinBuilder<'a> {
         let make_ctor_signature = |unifier: &mut Unifier| {
             unifier.add_ty(TypeEnum::TFunc(FunSignature {
                 args: vec![
-                    FuncArg { name: "start".into(), ty: int32, default_value: None },
+                    FuncArg {
+                        name: "start".into(),
+                        ty: int32,
+                        default_value: None,
+                        is_vararg: false,
+                    },
                     FuncArg {
                         name: "stop".into(),
                         ty: int32,
                         // placeholder
                         default_value: Some(SymbolValue::I32(0)),
+                        is_vararg: false,
                     },
                     FuncArg {
                         name: "step".into(),
                         ty: int32,
                         default_value: Some(SymbolValue::I32(1)),
+                        is_vararg: false,
                     },
                 ],
                 ret: range,
@@ -657,7 +710,7 @@ impl<'a> BuiltinBuilder<'a> {
                         let (zelf_ty, zelf) = obj.unwrap();
                         let zelf =
                             zelf.to_basic_value_enum(ctx, generator, zelf_ty)?.into_pointer_value();
-                        let zelf = RangeValue::from_ptr_val(zelf, Some("range"));
+                        let zelf = RangeValue::from_pointer_value(zelf, Some("range"));
 
                         let mut start = None;
                         let mut stop = None;
@@ -784,9 +837,9 @@ impl<'a> BuiltinBuilder<'a> {
             prim,
             &[
                 PrimDef::Option,
-                PrimDef::OptionIsSome,
-                PrimDef::OptionIsNone,
-                PrimDef::OptionUnwrap,
+                PrimDef::FunOptionIsSome,
+                PrimDef::FunOptionIsNone,
+                PrimDef::FunOptionUnwrap,
                 PrimDef::FunSome,
             ],
         );
@@ -799,9 +852,9 @@ impl<'a> BuiltinBuilder<'a> {
                 fields: Vec::default(),
                 attributes: Vec::default(),
                 methods: vec![
-                    Self::create_method(PrimDef::OptionIsSome, self.is_some_ty.0),
-                    Self::create_method(PrimDef::OptionIsNone, self.is_some_ty.0),
-                    Self::create_method(PrimDef::OptionUnwrap, self.unwrap_ty.0),
+                    Self::create_method(PrimDef::FunOptionIsSome, self.is_some_ty.0),
+                    Self::create_method(PrimDef::FunOptionIsNone, self.is_some_ty.0),
+                    Self::create_method(PrimDef::FunOptionUnwrap, self.unwrap_ty.0),
                 ],
                 ancestors: vec![TypeAnnotation::CustomClass {
                     id: prim.id(),
@@ -812,7 +865,7 @@ impl<'a> BuiltinBuilder<'a> {
                 loc: None,
             },
 
-            PrimDef::OptionUnwrap => TopLevelDef::Function {
+            PrimDef::FunOptionUnwrap => TopLevelDef::Function {
                 name: prim.name().into(),
                 simple_name: prim.simple_name().into(),
                 signature: self.unwrap_ty.0,
@@ -826,7 +879,7 @@ impl<'a> BuiltinBuilder<'a> {
                 loc: None,
             },
 
-            PrimDef::OptionIsNone | PrimDef::OptionIsSome => TopLevelDef::Function {
+            PrimDef::FunOptionIsNone | PrimDef::FunOptionIsSome => TopLevelDef::Function {
                 name: prim.name().to_string(),
                 simple_name: prim.simple_name().into(),
                 signature: self.is_some_ty.0,
@@ -847,10 +900,10 @@ impl<'a> BuiltinBuilder<'a> {
                         };
 
                         let returned_int = match prim {
-                            PrimDef::OptionIsNone => {
+                            PrimDef::FunOptionIsNone => {
                                 ctx.builder.build_is_null(ptr, prim.simple_name())
                             }
-                            PrimDef::OptionIsSome => {
+                            PrimDef::FunOptionIsSome => {
                                 ctx.builder.build_is_not_null(ptr, prim.simple_name())
                             }
                             _ => unreachable!(),
@@ -869,6 +922,7 @@ impl<'a> BuiltinBuilder<'a> {
                         name: "n".into(),
                         ty: self.option_tvar.ty,
                         default_value: None,
+                        is_vararg: false,
                     }],
                     ret: self.primitives.option,
                     vars: into_var_map([self.option_tvar]),
@@ -923,7 +977,7 @@ impl<'a> BuiltinBuilder<'a> {
     fn build_ndarray_class_related(&self, prim: PrimDef) -> TopLevelDef {
         debug_assert_prim_is_allowed(
             prim,
-            &[PrimDef::NDArray, PrimDef::NDArrayCopy, PrimDef::NDArrayFill],
+            &[PrimDef::NDArray, PrimDef::FunNDArrayCopy, PrimDef::FunNDArrayFill],
         );
 
         match prim {
@@ -934,8 +988,8 @@ impl<'a> BuiltinBuilder<'a> {
                 fields: Vec::default(),
                 attributes: Vec::default(),
                 methods: vec![
-                    Self::create_method(PrimDef::NDArrayCopy, self.ndarray_copy_ty.0),
-                    Self::create_method(PrimDef::NDArrayFill, self.ndarray_fill_ty.0),
+                    Self::create_method(PrimDef::FunNDArrayCopy, self.ndarray_copy_ty.0),
+                    Self::create_method(PrimDef::FunNDArrayFill, self.ndarray_fill_ty.0),
                 ],
                 ancestors: Vec::default(),
                 constructor: None,
@@ -943,7 +997,7 @@ impl<'a> BuiltinBuilder<'a> {
                 loc: None,
             },
 
-            PrimDef::NDArrayCopy => TopLevelDef::Function {
+            PrimDef::FunNDArrayCopy => TopLevelDef::Function {
                 name: prim.name().into(),
                 simple_name: prim.simple_name().into(),
                 signature: self.ndarray_copy_ty.0,
@@ -953,15 +1007,14 @@ impl<'a> BuiltinBuilder<'a> {
                 resolver: None,
                 codegen_callback: Some(Arc::new(GenCall::new(Box::new(
                     |ctx, obj, fun, args, generator| {
-                        todo!()
-                        // gen_ndarray_copy(ctx, &obj, fun, &args, generator)
-                        //     .map(|val| Some(val.as_basic_value_enum()))
+                        gen_ndarray_copy(ctx, &obj, fun, &args, generator)
+                            .map(|val| Some(val.as_basic_value_enum()))
                     },
                 )))),
                 loc: None,
             },
 
-            PrimDef::NDArrayFill => TopLevelDef::Function {
+            PrimDef::FunNDArrayFill => TopLevelDef::Function {
                 name: prim.name().into(),
                 simple_name: prim.simple_name().into(),
                 signature: self.ndarray_fill_ty.0,
@@ -971,9 +1024,8 @@ impl<'a> BuiltinBuilder<'a> {
                 resolver: None,
                 codegen_callback: Some(Arc::new(GenCall::new(Box::new(
                     |ctx, obj, fun, args, generator| {
-                        todo!()
-                        // gen_ndarray_fill(ctx, &obj, fun, &args, generator)?;
-                        // Ok(None)
+                        gen_ndarray_fill(ctx, &obj, fun, &args, generator)?;
+                        Ok(None)
                     },
                 )))),
                 loc: None,
@@ -1005,6 +1057,7 @@ impl<'a> BuiltinBuilder<'a> {
                     name: "n".into(),
                     ty: self.num_or_ndarray_ty.ty,
                     default_value: None,
+                    is_vararg: false,
                 }],
                 ret: self.num_or_ndarray_ty.ty,
                 vars: self.num_or_ndarray_var_map.clone(),
@@ -1053,7 +1106,7 @@ impl<'a> BuiltinBuilder<'a> {
         );
 
         // The size variant of the function determines the size of the returned int.
-        let int_sized = size_variant_to_int_type(size_variant, self.primitives);
+        let int_sized = size_variant.of_int(self.primitives);
 
         let ndarray_int_sized =
             make_ndarray_ty(self.unifier, self.primitives, Some(int_sized), Some(common_ndim.ty));
@@ -1078,7 +1131,7 @@ impl<'a> BuiltinBuilder<'a> {
                 let arg_ty = fun.0.args[0].ty;
                 let arg = args[0].1.clone().to_basic_value_enum(ctx, generator, arg_ty)?;
 
-                let ret_elem_ty = size_variant_to_int_type(size_variant, &ctx.primitives);
+                let ret_elem_ty = size_variant.of_int(&ctx.primitives);
                 Ok(Some(builtin_fns::call_round(generator, ctx, (arg_ty, arg), ret_elem_ty)?))
             }),
         )
@@ -1119,7 +1172,7 @@ impl<'a> BuiltinBuilder<'a> {
             make_ndarray_ty(self.unifier, self.primitives, Some(float), Some(common_ndim.ty));
 
         // The size variant of the function determines the type of int returned
-        let int_sized = size_variant_to_int_type(size_variant, self.primitives);
+        let int_sized = size_variant.of_int(self.primitives);
         let ndarray_int_sized =
             make_ndarray_ty(self.unifier, self.primitives, Some(int_sized), Some(common_ndim.ty));
 
@@ -1142,7 +1195,7 @@ impl<'a> BuiltinBuilder<'a> {
                 let arg_ty = fun.0.args[0].ty;
                 let arg = args[0].1.clone().to_basic_value_enum(ctx, generator, arg_ty)?;
 
-                let ret_elem_ty = size_variant_to_int_type(size_variant, &ctx.primitives);
+                let ret_elem_ty = size_variant.of_int(&ctx.primitives);
                 let func = match kind {
                     Kind::Ceil => builtin_fns::call_ceil,
                     Kind::Floor => builtin_fns::call_floor,
@@ -1193,14 +1246,13 @@ impl<'a> BuiltinBuilder<'a> {
             self.ndarray_float,
             &[(self.ndarray_factory_fn_shape_arg_tvar.ty, "shape")],
             Box::new(move |ctx, obj, fun, args, generator| {
-                todo!()
-                // let func = match prim {
-                //     PrimDef::FunNpNDArray | PrimDef::FunNpEmpty => gen_ndarray_empty,
-                //     PrimDef::FunNpZeros => gen_ndarray_zeros,
-                //     PrimDef::FunNpOnes => gen_ndarray_ones,
-                //     _ => unreachable!(),
-                // };
-                // func(ctx, &obj, fun, &args, generator).map(|val| Some(val.as_basic_value_enum()))
+                let func = match prim {
+                    PrimDef::FunNpNDArray | PrimDef::FunNpEmpty => gen_ndarray_empty,
+                    PrimDef::FunNpZeros => gen_ndarray_zeros,
+                    PrimDef::FunNpOnes => gen_ndarray_ones,
+                    _ => unreachable!(),
+                };
+                func(ctx, &obj, fun, &args, generator).map(|val| Some(val.as_basic_value_enum()))
             }),
         )
     }
@@ -1225,16 +1277,23 @@ impl<'a> BuiltinBuilder<'a> {
                     simple_name: prim.simple_name().into(),
                     signature: self.unifier.add_ty(TypeEnum::TFunc(FunSignature {
                         args: vec![
-                            FuncArg { name: "object".into(), ty: tv.ty, default_value: None },
+                            FuncArg {
+                                name: "object".into(),
+                                ty: tv.ty,
+                                default_value: None,
+                                is_vararg: false,
+                            },
                             FuncArg {
                                 name: "copy".into(),
                                 ty: bool,
                                 default_value: Some(SymbolValue::Bool(true)),
+                                is_vararg: false,
                             },
                             FuncArg {
                                 name: "ndmin".into(),
                                 ty: int32,
                                 default_value: Some(SymbolValue::U32(0)),
+                                is_vararg: false,
                             },
                         ],
                         ret: ndarray,
@@ -1246,9 +1305,8 @@ impl<'a> BuiltinBuilder<'a> {
                     resolver: None,
                     codegen_callback: Some(Arc::new(GenCall::new(Box::new(
                         |ctx, obj, fun, args, generator| {
-                            todo!()
-                            // gen_ndarray_array(ctx, &obj, fun, &args, generator)
-                            //     .map(|val| Some(val.as_basic_value_enum()))
+                            gen_ndarray_array(ctx, &obj, fun, &args, generator)
+                                .map(|val| Some(val.as_basic_value_enum()))
                         },
                     )))),
                     loc: None,
@@ -1266,9 +1324,8 @@ impl<'a> BuiltinBuilder<'a> {
                     // type variable
                     &[(self.list_int32, "shape"), (tv.ty, "fill_value")],
                     Box::new(move |ctx, obj, fun, args, generator| {
-                        todo!()
-                        // gen_ndarray_full(ctx, &obj, fun, &args, generator)
-                        //     .map(|val| Some(val.as_basic_value_enum()))
+                        gen_ndarray_full(ctx, &obj, fun, &args, generator)
+                            .map(|val| Some(val.as_basic_value_enum()))
                     }),
                 )
             }
@@ -1278,17 +1335,24 @@ impl<'a> BuiltinBuilder<'a> {
                     simple_name: prim.simple_name().into(),
                     signature: self.unifier.add_ty(TypeEnum::TFunc(FunSignature {
                         args: vec![
-                            FuncArg { name: "N".into(), ty: int32, default_value: None },
+                            FuncArg {
+                                name: "N".into(),
+                                ty: int32,
+                                default_value: None,
+                                is_vararg: false,
+                            },
                             // TODO(Derppening): Default values current do not work?
                             FuncArg {
                                 name: "M".into(),
                                 ty: int32,
                                 default_value: Some(SymbolValue::OptionNone),
+                                is_vararg: false,
                             },
                             FuncArg {
                                 name: "k".into(),
                                 ty: int32,
                                 default_value: Some(SymbolValue::I32(0)),
+                                is_vararg: false,
                             },
                         ],
                         ret: self.ndarray_float_2d,
@@ -1300,9 +1364,8 @@ impl<'a> BuiltinBuilder<'a> {
                     resolver: None,
                     codegen_callback: Some(Arc::new(GenCall::new(Box::new(
                         |ctx, obj, fun, args, generator| {
-                            todo!()
-                            // gen_ndarray_eye(ctx, &obj, fun, &args, generator)
-                            //     .map(|val| Some(val.as_basic_value_enum()))
+                            gen_ndarray_eye(ctx, &obj, fun, &args, generator)
+                                .map(|val| Some(val.as_basic_value_enum()))
                         },
                     )))),
                     loc: None,
@@ -1315,9 +1378,8 @@ impl<'a> BuiltinBuilder<'a> {
                 self.ndarray_float_2d,
                 &[(int32, "n")],
                 Box::new(|ctx, obj, fun, args, generator| {
-                    todo!()
-                    // gen_ndarray_identity(ctx, &obj, fun, &args, generator)
-                    //     .map(|val| Some(val.as_basic_value_enum()))
+                    gen_ndarray_identity(ctx, &obj, fun, &args, generator)
+                        .map(|val| Some(val.as_basic_value_enum()))
                 }),
             ),
             _ => unreachable!(),
@@ -1334,7 +1396,12 @@ impl<'a> BuiltinBuilder<'a> {
             name: prim.name().into(),
             simple_name: prim.simple_name().into(),
             signature: self.unifier.add_ty(TypeEnum::TFunc(FunSignature {
-                args: vec![FuncArg { name: "s".into(), ty: str, default_value: None }],
+                args: vec![FuncArg {
+                    name: "s".into(),
+                    ty: str,
+                    default_value: None,
+                    is_vararg: false,
+                }],
                 ret: str,
                 vars: VarMap::default(),
             })),
@@ -1398,31 +1465,21 @@ impl<'a> BuiltinBuilder<'a> {
     fn build_len_function(&mut self) -> TopLevelDef {
         let prim = PrimDef::FunLen;
 
-        let PrimitiveStore { uint64, int32, .. } = *self.primitives;
+        // Type handled in [`Inferencer::try_fold_special_call`]
+        let arg_tvar = self.unifier.get_dummy_var();
 
-        let tvar = self.unifier.get_fresh_var(Some("L".into()), None);
-        let list = self
-            .unifier
-            .subst(
-                self.primitives.list,
-                &into_var_map([TypeVar { id: self.list_tvar.id, ty: tvar.ty }]),
-            )
-            .unwrap();
-        let ndims = self.unifier.get_fresh_const_generic_var(uint64, Some("N".into()), None);
-        let ndarray = make_ndarray_ty(self.unifier, self.primitives, Some(tvar.ty), Some(ndims.ty));
-
-        let arg_ty = self.unifier.get_fresh_var_with_range(
-            &[list, ndarray, self.primitives.range],
-            Some("I".into()),
-            None,
-        );
         TopLevelDef::Function {
             name: prim.name().into(),
             simple_name: prim.simple_name().into(),
             signature: self.unifier.add_ty(TypeEnum::TFunc(FunSignature {
-                args: vec![FuncArg { name: "ls".into(), ty: arg_ty.ty, default_value: None }],
-                ret: int32,
-                vars: into_var_map([tvar, arg_ty]),
+                args: vec![FuncArg {
+                    name: "obj".into(),
+                    ty: arg_tvar.ty,
+                    default_value: None,
+                    is_vararg: false,
+                }],
+                ret: self.primitives.int32,
+                vars: into_var_map([arg_tvar]),
             })),
             var_id: Vec::default(),
             instance_to_symbol: HashMap::default(),
@@ -1430,86 +1487,10 @@ impl<'a> BuiltinBuilder<'a> {
             resolver: None,
             codegen_callback: Some(Arc::new(GenCall::new(Box::new(
                 move |ctx, _, fun, args, generator| {
-                    let range_ty = ctx.primitives.range;
                     let arg_ty = fun.0.args[0].ty;
                     let arg = args[0].1.clone().to_basic_value_enum(ctx, generator, arg_ty)?;
-                    Ok(if ctx.unifier.unioned(arg_ty, range_ty) {
-                        let arg = RangeValue::from_ptr_val(arg.into_pointer_value(), Some("range"));
-                        let (start, end, step) = destructure_range(ctx, arg);
-                        Some(calculate_len_for_slice_range(generator, ctx, start, end, step).into())
-                    } else {
-                        match &*ctx.unifier.get_ty_immutable(arg_ty) {
-                            TypeEnum::TObj { obj_id, .. } if *obj_id == PrimDef::List.id() => {
-                                let int32 = ctx.ctx.i32_type();
-                                let zero = int32.const_zero();
-                                let len = ctx
-                                    .build_gep_and_load(
-                                        arg.into_pointer_value(),
-                                        &[zero, int32.const_int(1, false)],
-                                        None,
-                                    )
-                                    .into_int_value();
-                                if len.get_type().get_bit_width() == 32 {
-                                    Some(len.into())
-                                } else {
-                                    Some(
-                                        ctx.builder
-                                            .build_int_truncate(len, int32, "len2i32")
-                                            .map(Into::into)
-                                            .unwrap(),
-                                    )
-                                }
-                            }
-                            TypeEnum::TObj { obj_id, .. } if *obj_id == PrimDef::NDArray.id() => {
-                                let llvm_i32 = ctx.ctx.i32_type();
-                                let llvm_usize = generator.get_size_type(ctx.ctx);
 
-                                let arg = NDArrayValue::from_ptr_val(
-                                    arg.into_pointer_value(),
-                                    llvm_usize,
-                                    None,
-                                );
-
-                                let ndims = arg.dim_sizes().size(ctx, generator);
-                                ctx.make_assert(
-                                    generator,
-                                    ctx.builder
-                                        .build_int_compare(
-                                            IntPredicate::NE,
-                                            ndims,
-                                            llvm_usize.const_zero(),
-                                            "",
-                                        )
-                                        .unwrap(),
-                                    "0:TypeError",
-                                    &format!("{name}() of unsized object", name = prim.name()),
-                                    [None, None, None],
-                                    ctx.current_loc,
-                                );
-
-                                let len = unsafe {
-                                    arg.dim_sizes().get_typed_unchecked(
-                                        ctx,
-                                        generator,
-                                        &llvm_usize.const_zero(),
-                                        None,
-                                    )
-                                };
-
-                                if len.get_type().get_bit_width() == 32 {
-                                    Some(len.into())
-                                } else {
-                                    Some(
-                                        ctx.builder
-                                            .build_int_truncate(len, llvm_i32, "len")
-                                            .map(Into::into)
-                                            .unwrap(),
-                                    )
-                                }
-                            }
-                            _ => unreachable!(),
-                        }
-                    })
+                    builtin_fns::call_len(generator, ctx, (arg_ty, arg)).map(|ret| Some(ret.into()))
                 },
             )))),
             loc: None,
@@ -1525,8 +1506,18 @@ impl<'a> BuiltinBuilder<'a> {
             simple_name: prim.simple_name().into(),
             signature: self.unifier.add_ty(TypeEnum::TFunc(FunSignature {
                 args: vec![
-                    FuncArg { name: "m".into(), ty: self.num_ty.ty, default_value: None },
-                    FuncArg { name: "n".into(), ty: self.num_ty.ty, default_value: None },
+                    FuncArg {
+                        name: "m".into(),
+                        ty: self.num_ty.ty,
+                        default_value: None,
+                        is_vararg: false,
+                    },
+                    FuncArg {
+                        name: "n".into(),
+                        ty: self.num_ty.ty,
+                        default_value: None,
+                        is_vararg: false,
+                    },
                 ],
                 ret: self.num_ty.ty,
                 vars: self.num_var_map.clone(),
@@ -1554,10 +1545,19 @@ impl<'a> BuiltinBuilder<'a> {
         }
     }
 
-    /// Build the functions `np_min()` and `np_max()`.
-    fn build_np_min_max_function(&mut self, prim: PrimDef) -> TopLevelDef {
-        debug_assert_prim_is_allowed(prim, &[PrimDef::FunNpMin, PrimDef::FunNpMax]);
+    /// Build the functions `np_max()`, `np_min()`, `np_argmax()` and `np_argmin()`
+    /// Calls `call_numpy_max_min` with the function name
+    fn build_np_max_min_function(&mut self, prim: PrimDef) -> TopLevelDef {
+        debug_assert_prim_is_allowed(
+            prim,
+            &[PrimDef::FunNpArgmin, PrimDef::FunNpArgmax, PrimDef::FunNpMin, PrimDef::FunNpMax],
+        );
 
+        let (var_map, ret_ty) = match prim {
+            PrimDef::FunNpArgmax | PrimDef::FunNpArgmin => {
+                (self.num_or_ndarray_var_map.clone(), self.primitives.int64)
+            }
+            PrimDef::FunNpMax | PrimDef::FunNpMin => {
                 let ret_ty = self.unifier.get_fresh_var(Some("R".into()), None);
                 let var_map = self
                     .num_or_ndarray_var_map
@@ -1565,28 +1565,25 @@ impl<'a> BuiltinBuilder<'a> {
                     .into_iter()
                     .chain(once((ret_ty.id, ret_ty.ty)))
                     .collect::<IndexMap<_, _>>();
+                (var_map, ret_ty.ty)
+            }
+            _ => unreachable!(),
+        };
 
         create_fn_by_codegen(
             self.unifier,
             &var_map,
             prim.name(),
-            ret_ty.ty,
-            &[(self.float_or_ndarray_ty.ty, "a")],
+            ret_ty,
+            &[(self.num_or_ndarray_ty.ty, "a")],
             Box::new(move |ctx, _, fun, args, generator| {
                 let a_ty = fun.0.args[0].ty;
                 let a = args[0].1.clone().to_basic_value_enum(ctx, generator, a_ty)?;
 
-                let func = match prim {
-                    PrimDef::FunNpMin => builtin_fns::call_numpy_min,
-                    PrimDef::FunNpMax => builtin_fns::call_numpy_max,
-                    _ => unreachable!(),
-                };
-
-                Ok(Some(func(generator, ctx, (a_ty, a))?))
+                Ok(Some(builtin_fns::call_numpy_max_min(generator, ctx, (a_ty, a), prim.name())?))
             }),
         )
     }
-
     /// Build the functions `np_minimum()` and `np_maximum()`.
     fn build_np_minimum_maximum_function(&mut self, prim: PrimDef) -> TopLevelDef {
         debug_assert_prim_is_allowed(prim, &[PrimDef::FunNpMinimum, PrimDef::FunNpMaximum]);
@@ -1602,7 +1599,12 @@ impl<'a> BuiltinBuilder<'a> {
             signature: self.unifier.add_ty(TypeEnum::TFunc(FunSignature {
                 args: param_ty
                     .iter()
-                    .map(|p| FuncArg { name: p.1.into(), ty: p.0, default_value: None })
+                    .map(|p| FuncArg {
+                        name: p.1.into(),
+                        ty: p.0,
+                        default_value: None,
+                        is_vararg: false,
+                    })
                     .collect(),
                 ret: ret_ty.ty,
                 vars: into_var_map([x1_ty, x2_ty, ret_ty]),
@@ -1643,6 +1645,7 @@ impl<'a> BuiltinBuilder<'a> {
                     name: "n".into(),
                     ty: self.num_or_ndarray_ty.ty,
                     default_value: None,
+                    is_vararg: false,
                 }],
                 ret: self.num_or_ndarray_ty.ty,
                 vars: self.num_or_ndarray_var_map.clone(),
@@ -1831,7 +1834,12 @@ impl<'a> BuiltinBuilder<'a> {
             signature: self.unifier.add_ty(TypeEnum::TFunc(FunSignature {
                 args: param_ty
                     .iter()
-                    .map(|p| FuncArg { name: p.1.into(), ty: p.0, default_value: None })
+                    .map(|p| FuncArg {
+                        name: p.1.into(),
+                        ty: p.0,
+                        default_value: None,
+                        is_vararg: false,
+                    })
                     .collect(),
                 ret: ret_ty.ty,
                 vars: into_var_map([x1_ty, x2_ty, ret_ty]),
@@ -1865,6 +1873,207 @@ impl<'a> BuiltinBuilder<'a> {
         }
     }
 
+    /// Build np/sp functions that take as input `NDArray` only
+    fn build_np_sp_ndarray_function(&mut self, prim: PrimDef) -> TopLevelDef {
+        debug_assert_prim_is_allowed(prim, &[PrimDef::FunNpTranspose, PrimDef::FunNpReshape]);
+
+        match prim {
+            PrimDef::FunNpTranspose => {
+                let ndarray_ty = self.unifier.get_fresh_var_with_range(
+                    &[self.ndarray_num_ty],
+                    Some("T".into()),
+                    None,
+                );
+                create_fn_by_codegen(
+                    self.unifier,
+                    &into_var_map([ndarray_ty]),
+                    prim.name(),
+                    ndarray_ty.ty,
+                    &[(ndarray_ty.ty, "x")],
+                    Box::new(move |ctx, _, fun, args, generator| {
+                        let arg_ty = fun.0.args[0].ty;
+                        let arg_val =
+                            args[0].1.clone().to_basic_value_enum(ctx, generator, arg_ty)?;
+                        Ok(Some(ndarray_transpose(generator, ctx, (arg_ty, arg_val))?))
+                    }),
+                )
+            }
+
+            // NOTE: on `ndarray_factory_fn_shape_arg_tvar` and
+            // the `param_ty` for `create_fn_by_codegen`.
+            //
+            // Similar to `build_ndarray_from_shape_factory_function` we delegate the responsibility of typechecking
+            // to [`typecheck::type_inferencer::Inferencer::fold_numpy_function_call_shape_argument`],
+            // and use a dummy [`TypeVar`] `ndarray_factory_fn_shape_arg_tvar` as a placeholder for `param_ty`.
+            PrimDef::FunNpReshape => create_fn_by_codegen(
+                self.unifier,
+                &VarMap::new(),
+                prim.name(),
+                self.ndarray_num_ty,
+                &[(self.ndarray_num_ty, "x"), (self.ndarray_factory_fn_shape_arg_tvar.ty, "shape")],
+                Box::new(move |ctx, _, fun, args, generator| {
+                    let x1_ty = fun.0.args[0].ty;
+                    let x1_val = args[0].1.clone().to_basic_value_enum(ctx, generator, x1_ty)?;
+                    let x2_ty = fun.0.args[1].ty;
+                    let x2_val = args[1].1.clone().to_basic_value_enum(ctx, generator, x2_ty)?;
+                    Ok(Some(ndarray_reshape(generator, ctx, (x1_ty, x1_val), (x2_ty, x2_val))?))
+                }),
+            ),
+
+            _ => unreachable!(),
+        }
+    }
+
+    /// Build `np_linalg` and `sp_linalg` functions
+    ///
+    /// The input to these functions must be floating point `NDArray`
+    fn build_linalg_methods(&mut self, prim: PrimDef) -> TopLevelDef {
+        debug_assert_prim_is_allowed(
+            prim,
+            &[
+                PrimDef::FunNpDot,
+                PrimDef::FunNpLinalgCholesky,
+                PrimDef::FunNpLinalgQr,
+                PrimDef::FunNpLinalgSvd,
+                PrimDef::FunNpLinalgInv,
+                PrimDef::FunNpLinalgPinv,
+                PrimDef::FunNpLinalgMatrixPower,
+                PrimDef::FunNpLinalgDet,
+                PrimDef::FunSpLinalgLu,
+                PrimDef::FunSpLinalgSchur,
+                PrimDef::FunSpLinalgHessenberg,
+            ],
+        );
+
+        match prim {
+            PrimDef::FunNpDot => create_fn_by_codegen(
+                self.unifier,
+                &self.num_or_ndarray_var_map,
+                prim.name(),
+                self.num_ty.ty,
+                &[(self.num_or_ndarray_ty.ty, "x1"), (self.num_or_ndarray_ty.ty, "x2")],
+                Box::new(move |ctx, _, fun, args, generator| {
+                    let x1_ty = fun.0.args[0].ty;
+                    let x1_val = args[0].1.clone().to_basic_value_enum(ctx, generator, x1_ty)?;
+                    let x2_ty = fun.0.args[1].ty;
+                    let x2_val = args[1].1.clone().to_basic_value_enum(ctx, generator, x2_ty)?;
+
+                    Ok(Some(ndarray_dot(generator, ctx, (x1_ty, x1_val), (x2_ty, x2_val))?))
+                }),
+            ),
+
+            PrimDef::FunNpLinalgCholesky | PrimDef::FunNpLinalgInv | PrimDef::FunNpLinalgPinv => {
+                create_fn_by_codegen(
+                    self.unifier,
+                    &VarMap::new(),
+                    prim.name(),
+                    self.ndarray_float_2d,
+                    &[(self.ndarray_float_2d, "x1")],
+                    Box::new(move |ctx, _, fun, args, generator| {
+                        let x1_ty = fun.0.args[0].ty;
+                        let x1_val =
+                            args[0].1.clone().to_basic_value_enum(ctx, generator, x1_ty)?;
+
+                        let func = match prim {
+                            PrimDef::FunNpLinalgCholesky => builtin_fns::call_np_linalg_cholesky,
+                            PrimDef::FunNpLinalgInv => builtin_fns::call_np_linalg_inv,
+                            PrimDef::FunNpLinalgPinv => builtin_fns::call_np_linalg_pinv,
+                            _ => unreachable!(),
+                        };
+                        Ok(Some(func(generator, ctx, (x1_ty, x1_val))?))
+                    }),
+                )
+            }
+
+            PrimDef::FunNpLinalgQr
+            | PrimDef::FunSpLinalgLu
+            | PrimDef::FunSpLinalgSchur
+            | PrimDef::FunSpLinalgHessenberg => {
+                let ret_ty = self.unifier.add_ty(TypeEnum::TTuple {
+                    ty: vec![self.ndarray_float_2d, self.ndarray_float_2d],
+                    is_vararg_ctx: false,
+                });
+                create_fn_by_codegen(
+                    self.unifier,
+                    &VarMap::new(),
+                    prim.name(),
+                    ret_ty,
+                    &[(self.ndarray_float_2d, "x1")],
+                    Box::new(move |ctx, _, fun, args, generator| {
+                        let x1_ty = fun.0.args[0].ty;
+                        let x1_val =
+                            args[0].1.clone().to_basic_value_enum(ctx, generator, x1_ty)?;
+
+                        let func = match prim {
+                            PrimDef::FunNpLinalgQr => builtin_fns::call_np_linalg_qr,
+                            PrimDef::FunSpLinalgLu => builtin_fns::call_sp_linalg_lu,
+                            PrimDef::FunSpLinalgSchur => builtin_fns::call_sp_linalg_schur,
+                            PrimDef::FunSpLinalgHessenberg => {
+                                builtin_fns::call_sp_linalg_hessenberg
+                            }
+                            _ => unreachable!(),
+                        };
+                        Ok(Some(func(generator, ctx, (x1_ty, x1_val))?))
+                    }),
+                )
+            }
+
+            PrimDef::FunNpLinalgSvd => {
+                let ret_ty = self.unifier.add_ty(TypeEnum::TTuple {
+                    ty: vec![self.ndarray_float_2d, self.ndarray_float, self.ndarray_float_2d],
+                    is_vararg_ctx: false,
+                });
+                create_fn_by_codegen(
+                    self.unifier,
+                    &VarMap::new(),
+                    prim.name(),
+                    ret_ty,
+                    &[(self.ndarray_float_2d, "x1")],
+                    Box::new(move |ctx, _, fun, args, generator| {
+                        let x1_ty = fun.0.args[0].ty;
+                        let x1_val =
+                            args[0].1.clone().to_basic_value_enum(ctx, generator, x1_ty)?;
+
+                        Ok(Some(builtin_fns::call_np_linalg_svd(generator, ctx, (x1_ty, x1_val))?))
+                    }),
+                )
+            }
+            PrimDef::FunNpLinalgMatrixPower => create_fn_by_codegen(
+                self.unifier,
+                &VarMap::new(),
+                prim.name(),
+                self.ndarray_float_2d,
+                &[(self.ndarray_float_2d, "x1"), (self.primitives.int32, "power")],
+                Box::new(move |ctx, _, fun, args, generator| {
+                    let x1_ty = fun.0.args[0].ty;
+                    let x1_val = args[0].1.clone().to_basic_value_enum(ctx, generator, x1_ty)?;
+                    let x2_ty = fun.0.args[1].ty;
+                    let x2_val = args[1].1.clone().to_basic_value_enum(ctx, generator, x2_ty)?;
+
+                    Ok(Some(builtin_fns::call_np_linalg_matrix_power(
+                        generator,
+                        ctx,
+                        (x1_ty, x1_val),
+                        (x2_ty, x2_val),
+                    )?))
+                }),
+            ),
+            PrimDef::FunNpLinalgDet => create_fn_by_codegen(
+                self.unifier,
+                &VarMap::new(),
+                prim.name(),
+                self.primitives.float,
+                &[(self.ndarray_float_2d, "x1")],
+                Box::new(move |ctx, _, fun, args, generator| {
+                    let x1_ty = fun.0.args[0].ty;
+                    let x1_val = args[0].1.clone().to_basic_value_enum(ctx, generator, x1_ty)?;
+                    Ok(Some(builtin_fns::call_np_linalg_det(generator, ctx, (x1_ty, x1_val))?))
+                }),
+            ),
+            _ => unreachable!(),
+        }
+    }
+
     fn create_method(prim: PrimDef, method_ty: Type) -> (StrRef, Type, DefinitionId) {
         (prim.simple_name().into(), method_ty, prim.id())
     }

nac3core/src/toplevel/composer.rs

@@ -1,17 +1,17 @@
-use nac3parser::ast::fold::Fold;
 use std::rc::Rc;
 
+use nac3parser::ast::{fold::Fold, ExprKind, Ident};
+
+use super::*;
 use crate::{
     codegen::{expr::get_subst_key, stmt::exn_constructor},
     symbol_resolver::SymbolValue,
     typecheck::{
-        type_inferencer::{FunctionData, Inferencer},
+        type_inferencer::{FunctionData, IdentifierInfo, Inferencer},
         typedef::{TypeVar, VarMap},
     },
 };
 
-use super::*;
-
 pub struct ComposerConfig {
     pub kernel_ann: Option<&'static str>,
     pub kernel_invariant_ann: &'static str,
@@ -23,7 +23,7 @@ impl Default for ComposerConfig {
     }
 }
 
-type DefAst = (Arc<RwLock<TopLevelDef>>, Option<Stmt<()>>);
+pub type DefAst = (Arc<RwLock<TopLevelDef>>, Option<Stmt<()>>);
 pub struct TopLevelComposer {
     // list of top level definitions, same as top level context
     pub definition_ast_list: Vec<DefAst>,
@@ -44,12 +44,27 @@ pub struct TopLevelComposer {
     pub size_t: u32,
 }
 
+/// The specification for a builtin function, consisting of the function name, the function
+/// signature, and a [code generation callback][`GenCall`].
+pub type BuiltinFuncSpec = (StrRef, FunSignature, Arc<GenCall>);
+
+/// A function that creates a [`BuiltinFuncSpec`] using the provided [`PrimitiveStore`] and
+/// [`Unifier`].
+pub type BuiltinFuncCreator = dyn Fn(&PrimitiveStore, &mut Unifier) -> BuiltinFuncSpec;
+
 impl TopLevelComposer {
     /// return a composer and things to make a "primitive" symbol resolver, so that the symbol
-    /// resolver can later figure out primitive type definitions when passed a primitive type name
+    /// resolver can later figure out primitive tye definitions when passed a primitive type name
+    ///
+    /// `lateinit_builtins` are specifically for the ARTIQ module. Since the [`Unifier`] instance
+    /// used to create builtin functions do not persist until method compilation, any types
+    /// created (e.g. [`TypeEnum::TVar`]) also do not persist. Those functions should be instead put
+    /// in `lateinit_builtins`, where they will be instantiated with the [`Unifier`] instance used
+    /// for method compilation.
     #[must_use]
     pub fn new(
-        builtins: Vec<(StrRef, FunSignature, Arc<GenCall>)>,
+        builtins: Vec<BuiltinFuncSpec>,
+        lateinit_builtins: Vec<Box<BuiltinFuncCreator>>,
         core_config: ComposerConfig,
         size_t: u32,
     ) -> (Self, HashMap<StrRef, DefinitionId>, HashMap<StrRef, Type>) {
@@ -86,7 +101,8 @@ impl TopLevelComposer {
             .iter()
             .map(|def_ast| match *def_ast.0.read() {
                 TopLevelDef::Class { name, .. } => name.to_string(),
-                TopLevelDef::Function { simple_name, .. } => simple_name.to_string(),
+                TopLevelDef::Function { simple_name, .. }
+                | TopLevelDef::Variable { simple_name, .. } => simple_name.to_string(),
             })
             .collect_vec();
 
@@ -119,7 +135,13 @@ impl TopLevelComposer {
             }
         }
 
-        for (name, sig, codegen_callback) in builtins {
+        // Materialize lateinit_builtins, now that the unifier is ready
+        let lateinit_builtins = lateinit_builtins
+            .into_iter()
+            .map(|builtin| builtin(&primitives_ty, &mut unifier))
+            .collect_vec();
+
+        for (name, sig, codegen_callback) in builtins.into_iter().chain(lateinit_builtins) {
             let fun_sig = unifier.add_ty(TypeEnum::TFunc(sig));
             builtin_ty.insert(name, fun_sig);
             builtin_id.insert(name, DefinitionId(definition_ast_list.len()));
@@ -360,13 +382,87 @@ impl TopLevelComposer {
                 ))
             }
 
+            ast::StmtKind::Assign { .. } => {
+                // Assignment statements can assign to (and therefore create) more than one 
+                // variable, but this function only allows returning one set of symbol information.
+                // We want to avoid changing this to return a `Vec` of symbol info, as this would 
+                // require `iter().next().unwrap()` on every variable created from a non-Assign 
+                // statement.
+                //
+                // Make callers use `register_top_level_var` instead, as it provides more
+                // fine-grained control over which symbols to register, while also simplifying the
+                // usage of this function.
+                panic!("Registration of top-level Assign statements must use TopLevelComposer::register_top_level_var (at {})", ast.location);
+            }
+
+            ast::StmtKind::AnnAssign { target, annotation, .. } => {
+                let ExprKind::Name { id: name, .. } = target.node else {
+                    return Err(format!(
+                        "global variable declaration must be an identifier (at {})",
+                        target.location
+                    ));
+                };
+
+                self.register_top_level_var(
+                    name,
+                    Some(annotation.as_ref().clone()),
+                    resolver,
+                    mod_path,
+                    target.location,
+                )
+            }
+
             _ => Err(format!(
-                "registrations of constructs other than top level classes/functions are not supported (at {})",
+                "registrations of constructs other than top level classes/functions/variables are not supported (at {})",
                 ast.location
             )),
         }
     }
 
+    /// Registers a top-level variable with the given `name` into the composer.
+    ///
+    /// `annotation` - The type annotation of the top-level variable, or [`None`] if no type
+    /// annotation is provided.
+    /// `location` - The location of the top-level variable.
+    pub fn register_top_level_var(
+        &mut self,
+        name: Ident,
+        annotation: Option<Expr>,
+        resolver: Option<Arc<dyn SymbolResolver + Send + Sync>>,
+        mod_path: &str,
+        location: Location,
+    ) -> Result<(StrRef, DefinitionId, Option<Type>), String> {
+        if self.keyword_list.contains(&name) {
+            return Err(format!("cannot use keyword `{name}` as a class name (at {location})"));
+        }
+
+        let global_var_name =
+            if mod_path.is_empty() { name.to_string() } else { format!("{mod_path}.{name}") };
+
+        if !self.defined_names.insert(global_var_name.clone()) {
+            return Err(format!(
+                "global variable `{global_var_name}` defined twice (at {location})"
+            ));
+        }
+
+        let ty_to_be_unified = self.unifier.get_dummy_var().ty;
+        self.definition_ast_list.push((
+            RwLock::new(Self::make_top_level_variable_def(
+                global_var_name,
+                name,
+                // dummy here, unify with correct type later,
+                ty_to_be_unified,
+                annotation,
+                resolver,
+                Some(location),
+            ))
+            .into(),
+            None,
+        ));
+
+        Ok((name, DefinitionId(self.definition_ast_list.len() - 1), Some(ty_to_be_unified)))
+    }
+
     pub fn start_analysis(&mut self, inference: bool) -> Result<(), HashSet<String>> {
         self.analyze_top_level_class_type_var()?;
         self.analyze_top_level_class_bases()?;
@@ -375,6 +471,7 @@ impl TopLevelComposer {
         if inference {
             self.analyze_function_instance()?;
         }
+        self.analyze_top_level_variables()?;
         Ok(())
     }
 
@@ -412,7 +509,7 @@ impl TopLevelComposer {
                     // things like `class A(Generic[T, V, ImportedModule.T])` is not supported
                     // i.e. only simple names are allowed in the subscript
                     // should update the TopLevelDef::Class.typevars and the TypeEnum::TObj.params
-                    ast::ExprKind::Subscript { value, slice, .. }
+                    ExprKind::Subscript { value, slice, .. }
                         if {
                             matches!(
                                 &value.node,
@@ -428,9 +525,9 @@ impl TopLevelComposer {
                         }
                         is_generic = true;
 
-                        let type_var_list: Vec<&ast::Expr<()>>;
+                        let type_var_list: Vec<&Expr<()>>;
                         // if `class A(Generic[T, V, G])`
-                        if let ast::ExprKind::Tuple { elts, .. } = &slice.node {
+                        if let ExprKind::Tuple { elts, .. } = &slice.node {
                             type_var_list = elts.iter().collect_vec();
                         // `class A(Generic[T])`
                         } else {
@@ -479,6 +576,7 @@ impl TopLevelComposer {
             }
             Ok(())
         };
+
         let mut errors = HashSet::new();
         for (class_def, class_ast) in def_list.iter().skip(self.builtin_num) {
             if class_ast.is_none() {
@@ -766,6 +864,7 @@ impl TopLevelComposer {
             let target_ty = get_type_from_type_annotation_kinds(
                 &temp_def_list,
                 unifier,
+                primitives,
                 &def,
                 &mut subst_list,
             )?;
@@ -831,7 +930,6 @@ impl TopLevelComposer {
         let unifier = self.unifier.borrow_mut();
         let primitives_store = &self.primitives_ty;
 
-        let mut errors = HashSet::new();
         let mut analyze = |function_def: &Arc<RwLock<TopLevelDef>>, function_ast: &Option<Stmt>| {
             let mut function_def = function_def.write();
             let function_def = &mut *function_def;
@@ -859,7 +957,73 @@ impl TopLevelComposer {
             let resolver = &**resolver;
 
             let mut function_var_map = VarMap::new();
-            let arg_types = {
+
+            let vararg = args
+                .vararg
+                .as_ref()
+                .map(|vararg| -> Result<_, HashSet<String>> {
+                    let vararg = vararg.as_ref();
+
+                    let annotation = vararg
+                        .node
+                        .annotation
+                        .as_ref()
+                        .ok_or_else(|| {
+                            HashSet::from([format!(
+                                "function parameter `{}` needs type annotation at {}",
+                                vararg.node.arg, vararg.location
+                            )])
+                        })?
+                        .as_ref();
+
+                    let type_annotation = parse_ast_to_type_annotation_kinds(
+                        resolver,
+                        temp_def_list.as_slice(),
+                        unifier,
+                        primitives_store,
+                        annotation,
+                        // NOTE: since only class need this, for function
+                        // it should be fine to be empty map
+                        HashMap::new(),
+                    )?;
+
+                    let type_vars_within =
+                        get_type_var_contained_in_type_annotation(&type_annotation)
+                            .into_iter()
+                            .map(|x| -> Result<TypeVar, HashSet<String>> {
+                                let TypeAnnotation::TypeVar(ty) = x else {
+                                    unreachable!("must be type var annotation kind")
+                                };
+
+                                let id = Self::get_var_id(ty, unifier)?;
+                                Ok(TypeVar { id, ty })
+                            })
+                            .collect::<Result<Vec<_>, _>>()?;
+                    for var in type_vars_within {
+                        if let Some(prev_ty) = function_var_map.insert(var.id, var.ty) {
+                            // if already have the type inserted, make sure they are the same thing
+                            assert_eq!(prev_ty, var.ty);
+                        }
+                    }
+
+                    let ty = get_type_from_type_annotation_kinds(
+                        temp_def_list.as_ref(),
+                        unifier,
+                        primitives_store,
+                        &type_annotation,
+                        &mut None,
+                    )?;
+
+                    Ok(FuncArg {
+                        name: vararg.node.arg,
+                        ty,
+                        default_value: Some(SymbolValue::Tuple(Vec::default())),
+                        is_vararg: true,
+                    })
+                })
+                .transpose()?;
+
+            let mut arg_types = {
                 // make sure no duplicate parameter
                 let mut defined_parameter_name: HashSet<_> = HashSet::new();
                 for x in &args.args {
@@ -874,15 +1038,15 @@ impl TopLevelComposer {
                     }
                 }
 
-                let arg_with_default: Vec<(&ast::Located<ast::ArgData<()>>, Option<&ast::Expr>)> =
-                    args.args
+                let arg_with_default: Vec<(&ast::Located<ast::ArgData<()>>, Option<&Expr>)> = args
+                    .args
                     .iter()
                     .rev()
                     .zip(
                         args.defaults
                             .iter()
                             .rev()
-                                .map(|x| -> Option<&ast::Expr> { Some(x) })
+                            .map(|x| -> Option<&Expr> { Some(x) })
                             .chain(std::iter::repeat(None)),
                     )
                     .collect_vec();
@@ -936,6 +1100,7 @@ impl TopLevelComposer {
                         let ty = get_type_from_type_annotation_kinds(
                             temp_def_list.as_ref(),
                             unifier,
+                            primitives_store,
                             &type_annotation,
                             &mut None,
                         )?;
@@ -959,11 +1124,18 @@ impl TopLevelComposer {
                                     v
                                 }),
                             },
+                            is_vararg: false,
                         })
                     })
                     .collect::<Result<Vec<_>, _>>()?
             };
 
+            if let Some(vararg) = vararg {
+                arg_types.push(vararg);
+            };
+
+            let arg_types = arg_types;
+
             let return_ty = {
                 if let Some(returns) = returns {
                     let return_ty_annotation = {
@@ -1002,6 +1174,7 @@ impl TopLevelComposer {
                     get_type_from_type_annotation_kinds(
                         &temp_def_list,
                         unifier,
+                        primitives_store,
                         &return_ty_annotation,
                         &mut None,
                     )?
@@ -1031,6 +1204,8 @@ impl TopLevelComposer {
             })?;
             Ok(())
         };
+
+        let mut errors = HashSet::new();
         for (function_def, function_ast) in def_list.iter().skip(self.builtin_num) {
             if function_ast.is_none() {
                 continue;
@@ -1132,7 +1307,7 @@ impl TopLevelComposer {
 
                         let arg_with_default: Vec<(
                             &ast::Located<ast::ArgData<()>>,
-                            Option<&ast::Expr>,
+                            Option<&Expr>,
                         )> = args
                             .args
                             .iter()
@@ -1141,7 +1316,7 @@ impl TopLevelComposer {
                                 args.defaults
                                     .iter()
                                     .rev()
-                                    .map(|x| -> Option<&ast::Expr> { Some(x) })
+                                    .map(|x| -> Option<&Expr> { Some(x) })
                                     .chain(std::iter::repeat(None)),
                             )
                             .collect_vec();
@@ -1214,6 +1389,7 @@ impl TopLevelComposer {
                                             })
                                         }
                                     },
+                                    is_vararg: false,
                                 };
                                 // push the dummy type and the type annotation
                                 // into the list for later unification
@@ -1297,7 +1473,7 @@ impl TopLevelComposer {
                         .map_err(|e| HashSet::from([e.to_display(unifier).to_string()]))?;
                 }
                 ast::StmtKind::AnnAssign { target, annotation, value, .. } => {
-                    if let ast::ExprKind::Name { id: attr, .. } = &target.node {
+                    if let ExprKind::Name { id: attr, .. } = &target.node {
                         if defined_fields.insert(attr.to_string()) {
                             let dummy_field_type = unifier.get_dummy_var().ty;
 
@@ -1305,7 +1481,7 @@ impl TopLevelComposer {
                                 None => {
                                     // handle Kernel[T], KernelInvariant[T]
                                     let (annotation, mutable) = match &annotation.node {
-                                        ast::ExprKind::Subscript { value, slice, .. }
+                                        ExprKind::Subscript { value, slice, .. }
                                             if matches!(
                                                 &value.node,
                                                 ast::ExprKind::Name { id, .. } if id == &core_config.kernel_invariant_ann.into()
@@ -1313,7 +1489,7 @@ impl TopLevelComposer {
                                         {
                                             (slice, false)
                                         }
-                                        ast::ExprKind::Subscript { value, slice, .. }
+                                        ExprKind::Subscript { value, slice, .. }
                                             if matches!(
                                                 &value.node,
                                                 ast::ExprKind::Name { id, .. } if core_config.kernel_ann.map_or(false, |c| id == &c.into())
@@ -1331,13 +1507,13 @@ impl TopLevelComposer {
                                 Some(boxed_expr) => {
                                     // Class attributes are set as immutable regardless
                                     let (annotation, _) = match &annotation.node {
-                                        ast::ExprKind::Subscript { slice, .. } => (slice, false),
+                                        ExprKind::Subscript { slice, .. } => (slice, false),
                                         _ if core_config.kernel_ann.is_none() => (annotation, false),
                                         _ => continue,
                                     };
 
                                     match &**boxed_expr {
-                                        ast::Located {location: _, custom: (), node: ast::ExprKind::Constant { value: v, kind: _ }} => {
+                                        ast::Located {location: _, custom: (), node: ExprKind::Constant { value: v, kind: _ }} => {
                                             // Restricting the types allowed to be defined as class attributes
                                             match v {
                                                 ast::Constant::Bool(_) | ast::Constant::Str(_) | ast::Constant::Int(_) | ast::Constant::Float(_) => {}
@@ -1604,7 +1780,6 @@ impl TopLevelComposer {
             }
         }
 
-        let mut errors = HashSet::new();
         let mut analyze = |i, def: &Arc<RwLock<TopLevelDef>>, ast: &Option<Stmt>| {
             let class_def = def.read();
             if let TopLevelDef::Class {
@@ -1622,6 +1797,7 @@ impl TopLevelComposer {
                 let self_type = get_type_from_type_annotation_kinds(
                     &def_list,
                     unifier,
+                    primitives_ty,
                     &make_self_type_annotation(type_vars, *object_id),
                     &mut None,
                 )?;
@@ -1638,21 +1814,25 @@ impl TopLevelComposer {
                                 name: "msg".into(),
                                 ty: string,
                                 default_value: Some(SymbolValue::Str(String::new())),
+                                is_vararg: false,
                             },
                             FuncArg {
                                 name: "param0".into(),
                                 ty: int64,
                                 default_value: Some(SymbolValue::I64(0)),
+                                is_vararg: false,
                             },
                             FuncArg {
                                 name: "param1".into(),
                                 ty: int64,
                                 default_value: Some(SymbolValue::I64(0)),
+                                is_vararg: false,
                             },
                             FuncArg {
                                 name: "param2".into(),
                                 ty: int64,
                                 default_value: Some(SymbolValue::I64(0)),
+                                is_vararg: false,
                             },
                         ],
                         ret: self_type,
@@ -1719,7 +1899,12 @@ impl TopLevelComposer {
                         if *name != init_str_id {
                             unreachable!("must be init function here")
                         }
-                        let all_inited = Self::get_all_assigned_field(body.as_slice())?;
+
+                        let all_inited = Self::get_all_assigned_field(
+                            object_id.0,
+                            definition_ast_list,
+                            body.as_slice(),
+                        )?;
                         for (f, _, _) in fields {
                             if !all_inited.contains(f) {
                                 return Err(HashSet::from([
@@ -1737,6 +1922,8 @@ impl TopLevelComposer {
             }
             Ok(())
         };
+
+        let mut errors = HashSet::new();
         for (i, (def, ast)) in definition_ast_list.iter().enumerate().skip(self.builtin_num) {
             if ast.is_none() {
                 continue;
@@ -1774,19 +1961,20 @@ impl TopLevelComposer {
             if ast.is_none() {
                 return Ok(());
             }
-            let mut function_def = def.write();
-            if let TopLevelDef::Function {
-                instance_to_stmt,
-                instance_to_symbol,
-                name,
-                simple_name,
-                signature,
-                resolver,
-                ..
-            } = &mut *function_def
-            {
-                let signature_ty_enum = unifier.get_ty(*signature);
-                let TypeEnum::TFunc(FunSignature { args, ret, vars }) = signature_ty_enum.as_ref()
+
+            let (name, simple_name, signature, resolver) = {
+                let function_def = def.read();
+                let TopLevelDef::Function { name, simple_name, signature, resolver, .. } =
+                    &*function_def
+                else {
+                    return Ok(());
+                };
+
+                (name.clone(), *simple_name, *signature, resolver.clone())
+            };
+
+            let signature_ty_enum = unifier.get_ty(signature);
+            let TypeEnum::TFunc(FunSignature { args, ret, vars, .. }) = signature_ty_enum.as_ref()
             else {
                 unreachable!("must be typeenum::tfunc")
             };
@@ -1803,7 +1991,11 @@ impl TopLevelComposer {
 
                     let ty_ann = make_self_type_annotation(type_vars, *class_id);
                     let self_ty = get_type_from_type_annotation_kinds(
-                            &def_list, unifier, &ty_ann, &mut None,
+                        &def_list,
+                        unifier,
+                        primitives_ty,
+                        &ty_ann,
+                        &mut None,
                     )?;
                     vars.extend(type_vars.iter().map(|ty| {
                         let TypeEnum::TVar { id, .. } = &*unifier.get_ty(*ty) else {
@@ -1858,6 +2050,7 @@ impl TopLevelComposer {
                             name: a.name,
                             ty: unifier.subst(a.ty, &subst).unwrap_or(a.ty),
                             default_value: a.default_value.clone(),
+                            is_vararg: false,
                         })
                         .collect_vec()
                 };
@@ -1889,11 +2082,11 @@ impl TopLevelComposer {
                     })
                 };
                 let mut identifiers = {
-                        let mut result: HashSet<_> = HashSet::new();
+                    let mut result = HashMap::new();
                     if self_type.is_some() {
-                            result.insert("self".into());
+                        result.insert("self".into(), IdentifierInfo::default());
                     }
-                        result.extend(inst_args.iter().map(|x| x.name));
+                    result.extend(inst_args.iter().map(|x| (x.name, IdentifierInfo::default())));
                     result
                 };
                 let mut calls: HashMap<CodeLocation, CallId> = HashMap::new();
@@ -1930,23 +2123,43 @@ impl TopLevelComposer {
                 else {
                     unreachable!("must be function def ast")
                 };
-                    if !decorator_list.is_empty()
-                        && matches!(&decorator_list[0].node,
-                            ast::ExprKind::Name{ id, .. } if id == &"extern".into())
-                    {
-                        instance_to_symbol.insert(String::new(), simple_name.to_string());
-                        continue;
-                    }
-                    if !decorator_list.is_empty()
-                        && matches!(&decorator_list[0].node,
-                            ast::ExprKind::Name{ id, .. } if id == &"rpc".into())
+
+                if !decorator_list.is_empty() {
+                    if matches!(&decorator_list[0].node, ExprKind::Name { id, .. } if id == &"extern".into())
                     {
+                        let TopLevelDef::Function { instance_to_symbol, .. } = &mut *def.write()
+                        else {
+                            unreachable!()
+                        };
                         instance_to_symbol.insert(String::new(), simple_name.to_string());
                         continue;
                     }
 
-                    let fun_body = body
-                        .into_iter()
+                    if matches!(&decorator_list[0].node, ExprKind::Name { id, .. } if id == &"rpc".into())
+                    {
+                        let TopLevelDef::Function { instance_to_symbol, .. } = &mut *def.write()
+                        else {
+                            unreachable!()
+                        };
+                        instance_to_symbol.insert(String::new(), simple_name.to_string());
+                        continue;
+                    }
+
+                    if let ExprKind::Call { func, .. } = &decorator_list[0].node {
+                        if matches!(&func.node, ExprKind::Name { id, .. } if id == &"rpc".into()) {
+                            let TopLevelDef::Function { instance_to_symbol, .. } =
+                                &mut *def.write()
+                            else {
+                                unreachable!()
+                            };
+                            instance_to_symbol.insert(String::new(), simple_name.to_string());
+                            continue;
+                        }
+                    }
+                }
+
+                let fun_body =
+                    body.into_iter()
                         .map(|b| inferencer.fold_stmt(b))
                         .collect::<Result<Vec<_>, _>>()?;
 
@@ -2016,6 +2229,9 @@ impl TopLevelComposer {
                     )]));
                 }
 
+                let TopLevelDef::Function { instance_to_stmt, .. } = &mut *def.write() else {
+                    unreachable!()
+                };
                 instance_to_stmt.insert(
                     get_subst_key(
                         unifier,
@@ -2031,10 +2247,10 @@ impl TopLevelComposer {
                     },
                 );
             }
-            }
 
             Ok(())
         };
+
         for (id, (def, ast)) in self.definition_ast_list.iter().enumerate().skip(self.builtin_num) {
             if ast.is_none() {
                 continue;
@@ -2048,4 +2264,59 @@ impl TopLevelComposer {
         }
         Ok(())
     }
+
+    /// Step 6. Analyze and populate the types of global variables.
+    fn analyze_top_level_variables(&mut self) -> Result<(), HashSet<String>> {
+        let def_list = &self.definition_ast_list;
+        let temp_def_list = self.extract_def_list();
+        let unifier = &mut self.unifier;
+        let primitives_store = &self.primitives_ty;
+
+        let mut analyze = |variable_def: &Arc<RwLock<TopLevelDef>>| -> Result<_, HashSet<String>> {
+            let TopLevelDef::Variable { ty: dummy_ty, ty_decl, resolver, loc, .. } =
+                &*variable_def.read()
+            else {
+                // not top level variable def, skip
+                return Ok(());
+            };
+
+            let resolver = &**resolver.as_ref().unwrap();
+
+            if let Some(ty_decl) = ty_decl {
+                let ty_annotation = parse_ast_to_type_annotation_kinds(
+                    resolver,
+                    &temp_def_list,
+                    unifier,
+                    primitives_store,
+                    ty_decl,
+                    HashMap::new(),
+                )?;
+                let ty_from_ty_annotation = get_type_from_type_annotation_kinds(
+                    &temp_def_list,
+                    unifier,
+                    primitives_store,
+                    &ty_annotation,
+                    &mut None,
+                )?;
+
+                unifier.unify(*dummy_ty, ty_from_ty_annotation).map_err(|e| {
+                    HashSet::from([e.at(Some(loc.unwrap())).to_display(unifier).to_string()])
+                })?;
+            }
+
+            Ok(())
+        };
+
+        let mut errors = HashSet::new();
+        for (variable_def, _) in def_list.iter().skip(self.builtin_num) {
+            if let Err(e) = analyze(variable_def) {
+                errors.extend(e);
+            }
+        }
+        if !errors.is_empty() {
+            return Err(errors);
+        }
+
+        Ok(())
+    }
 }

nac3core/src/toplevel/helper.rs

@@ -1,13 +1,15 @@
 use std::convert::TryInto;
 
-use crate::symbol_resolver::SymbolValue;
-use crate::toplevel::numpy::unpack_ndarray_var_tys;
-use crate::typecheck::typedef::{into_var_map, iter_type_vars, Mapping, TypeVarId, VarMap};
-use nac3parser::ast::{Constant, Location};
 use strum::IntoEnumIterator;
 use strum_macros::EnumIter;
 
-use super::*;
+use nac3parser::ast::{Constant, ExprKind, Location};
+
+use super::{numpy::unpack_ndarray_var_tys, *};
+use crate::{
+    symbol_resolver::SymbolValue,
+    typecheck::typedef::{into_var_map, iter_type_vars, Mapping, TypeVarId, VarMap},
+};
 
 /// All primitive types and functions in nac3core.
 #[derive(Clone, Copy, Debug, EnumIter, PartialEq, Eq)]
@@ -27,17 +29,22 @@ pub enum PrimDef {
     List,
     NDArray,
 
-    // Member Functions
-    OptionIsSome,
-    OptionIsNone,
-    OptionUnwrap,
-    NDArrayCopy,
-    NDArrayFill,
-    FunInt32,
-    FunInt64,
-    FunUInt32,
-    FunUInt64,
-    FunFloat,
+    // Option methods
+    FunOptionIsSome,
+    FunOptionIsNone,
+    FunOptionUnwrap,
+
+    // Option-related functions
+    FunSome,
+
+    // NDArray methods
+    FunNDArrayCopy,
+    FunNDArrayFill,
+
+    // Range methods
+    FunRangeInit,
+
+    // NumPy factory functions
     FunNpNDArray,
     FunNpEmpty,
     FunNpZeros,
@@ -46,26 +53,17 @@ pub enum PrimDef {
     FunNpArray,
     FunNpEye,
     FunNpIdentity,
-    FunRound,
-    FunRound64,
+
+    // Miscellaneous NumPy & SciPy functions
     FunNpRound,
-    FunRangeInit,
-    FunStr,
-    FunBool,
-    FunFloor,
-    FunFloor64,
     FunNpFloor,
-    FunCeil,
-    FunCeil64,
     FunNpCeil,
-    FunLen,
-    FunMin,
     FunNpMin,
     FunNpMinimum,
-    FunMax,
+    FunNpArgmin,
     FunNpMax,
     FunNpMaximum,
-    FunAbs,
+    FunNpArgmax,
     FunNpIsNan,
     FunNpIsInf,
     FunNpSin,
@@ -103,15 +101,46 @@ pub enum PrimDef {
     FunNpLdExp,
     FunNpHypot,
     FunNpNextAfter,
+    FunNpTranspose,
+    FunNpReshape,
 
-    // Top-Level Functions
-    FunSome,
+    // Linalg functions
+    FunNpDot,
+    FunNpLinalgCholesky,
+    FunNpLinalgQr,
+    FunNpLinalgSvd,
+    FunNpLinalgInv,
+    FunNpLinalgPinv,
+    FunNpLinalgMatrixPower,
+    FunNpLinalgDet,
+    FunSpLinalgLu,
+    FunSpLinalgSchur,
+    FunSpLinalgHessenberg,
+
+    // Miscellaneous Python & NAC3 functions
+    FunInt32,
+    FunInt64,
+    FunUInt32,
+    FunUInt64,
+    FunFloat,
+    FunRound,
+    FunRound64,
+    FunStr,
+    FunBool,
+    FunFloor,
+    FunFloor64,
+    FunCeil,
+    FunCeil64,
+    FunLen,
+    FunMin,
+    FunMax,
+    FunAbs,
 }
 
 /// Associated details of a [`PrimDef`]
 pub enum PrimDefDetails {
     PrimFunction { name: &'static str, simple_name: &'static str },
-    PrimClass { name: &'static str },
+    PrimClass { name: &'static str, get_ty_fn: fn(&PrimitiveStore) -> Type },
 }
 
 impl PrimDef {
@@ -153,15 +182,17 @@ impl PrimDef {
     #[must_use]
     pub fn name(&self) -> &'static str {
         match self.details() {
-            PrimDefDetails::PrimFunction { name, .. } | PrimDefDetails::PrimClass { name } => name,
+            PrimDefDetails::PrimFunction { name, .. } | PrimDefDetails::PrimClass { name, .. } => {
+                name
+            }
         }
     }
 
     /// Get the associated details of this [`PrimDef`]
     #[must_use]
     pub fn details(self) -> PrimDefDetails {
-        fn class(name: &'static str) -> PrimDefDetails {
-            PrimDefDetails::PrimClass { name }
+        fn class(name: &'static str, get_ty_fn: fn(&PrimitiveStore) -> Type) -> PrimDefDetails {
+            PrimDefDetails::PrimClass { name, get_ty_fn }
         }
 
         fn fun(name: &'static str, simple_name: Option<&'static str>) -> PrimDefDetails {
@@ -169,29 +200,37 @@ impl PrimDef {
         }
 
         match self {
-            PrimDef::Int32 => class("int32"),
-            PrimDef::Int64 => class("int64"),
-            PrimDef::Float => class("float"),
-            PrimDef::Bool => class("bool"),
-            PrimDef::None => class("none"),
-            PrimDef::Range => class("range"),
-            PrimDef::Str => class("str"),
-            PrimDef::Exception => class("Exception"),
-            PrimDef::UInt32 => class("uint32"),
-            PrimDef::UInt64 => class("uint64"),
-            PrimDef::Option => class("Option"),
-            PrimDef::OptionIsSome => fun("Option.is_some", Some("is_some")),
-            PrimDef::OptionIsNone => fun("Option.is_none", Some("is_none")),
-            PrimDef::OptionUnwrap => fun("Option.unwrap", Some("unwrap")),
-            PrimDef::List => class("list"),
-            PrimDef::NDArray => class("ndarray"),
-            PrimDef::NDArrayCopy => fun("ndarray.copy", Some("copy")),
-            PrimDef::NDArrayFill => fun("ndarray.fill", Some("fill")),
-            PrimDef::FunInt32 => fun("int32", None),
-            PrimDef::FunInt64 => fun("int64", None),
-            PrimDef::FunUInt32 => fun("uint32", None),
-            PrimDef::FunUInt64 => fun("uint64", None),
-            PrimDef::FunFloat => fun("float", None),
+            // Classes
+            PrimDef::Int32 => class("int32", |primitives| primitives.int32),
+            PrimDef::Int64 => class("int64", |primitives| primitives.int64),
+            PrimDef::Float => class("float", |primitives| primitives.float),
+            PrimDef::Bool => class("bool", |primitives| primitives.bool),
+            PrimDef::None => class("none", |primitives| primitives.none),
+            PrimDef::Range => class("range", |primitives| primitives.range),
+            PrimDef::Str => class("str", |primitives| primitives.str),
+            PrimDef::Exception => class("Exception", |primitives| primitives.exception),
+            PrimDef::UInt32 => class("uint32", |primitives| primitives.uint32),
+            PrimDef::UInt64 => class("uint64", |primitives| primitives.uint64),
+            PrimDef::Option => class("Option", |primitives| primitives.option),
+            PrimDef::List => class("list", |primitives| primitives.list),
+            PrimDef::NDArray => class("ndarray", |primitives| primitives.ndarray),
+
+            // Option methods
+            PrimDef::FunOptionIsSome => fun("Option.is_some", Some("is_some")),
+            PrimDef::FunOptionIsNone => fun("Option.is_none", Some("is_none")),
+            PrimDef::FunOptionUnwrap => fun("Option.unwrap", Some("unwrap")),
+
+            // Option-related functions
+            PrimDef::FunSome => fun("Some", None),
+
+            // NDArray methods
+            PrimDef::FunNDArrayCopy => fun("ndarray.copy", Some("copy")),
+            PrimDef::FunNDArrayFill => fun("ndarray.fill", Some("fill")),
+
+            // Range methods
+            PrimDef::FunRangeInit => fun("range.__init__", Some("__init__")),
+
+            // NumPy factory functions
             PrimDef::FunNpNDArray => fun("np_ndarray", None),
             PrimDef::FunNpEmpty => fun("np_empty", None),
             PrimDef::FunNpZeros => fun("np_zeros", None),
@@ -200,26 +239,17 @@ impl PrimDef {
             PrimDef::FunNpArray => fun("np_array", None),
             PrimDef::FunNpEye => fun("np_eye", None),
             PrimDef::FunNpIdentity => fun("np_identity", None),
-            PrimDef::FunRound => fun("round", None),
-            PrimDef::FunRound64 => fun("round64", None),
+
+            // Miscellaneous NumPy & SciPy functions
             PrimDef::FunNpRound => fun("np_round", None),
-            PrimDef::FunRangeInit => fun("range.__init__", Some("__init__")),
-            PrimDef::FunStr => fun("str", None),
-            PrimDef::FunBool => fun("bool", None),
-            PrimDef::FunFloor => fun("floor", None),
-            PrimDef::FunFloor64 => fun("floor64", None),
             PrimDef::FunNpFloor => fun("np_floor", None),
-            PrimDef::FunCeil => fun("ceil", None),
-            PrimDef::FunCeil64 => fun("ceil64", None),
             PrimDef::FunNpCeil => fun("np_ceil", None),
-            PrimDef::FunLen => fun("len", None),
-            PrimDef::FunMin => fun("min", None),
             PrimDef::FunNpMin => fun("np_min", None),
             PrimDef::FunNpMinimum => fun("np_minimum", None),
-            PrimDef::FunMax => fun("max", None),
+            PrimDef::FunNpArgmin => fun("np_argmin", None),
             PrimDef::FunNpMax => fun("np_max", None),
             PrimDef::FunNpMaximum => fun("np_maximum", None),
-            PrimDef::FunAbs => fun("abs", None),
+            PrimDef::FunNpArgmax => fun("np_argmax", None),
             PrimDef::FunNpIsNan => fun("np_isnan", None),
             PrimDef::FunNpIsInf => fun("np_isinf", None),
             PrimDef::FunNpSin => fun("np_sin", None),
@@ -257,7 +287,40 @@ impl PrimDef {
             PrimDef::FunNpLdExp => fun("np_ldexp", None),
             PrimDef::FunNpHypot => fun("np_hypot", None),
             PrimDef::FunNpNextAfter => fun("np_nextafter", None),
-            PrimDef::FunSome => fun("Some", None),
+            PrimDef::FunNpTranspose => fun("np_transpose", None),
+            PrimDef::FunNpReshape => fun("np_reshape", None),
+
+            // Linalg functions
+            PrimDef::FunNpDot => fun("np_dot", None),
+            PrimDef::FunNpLinalgCholesky => fun("np_linalg_cholesky", None),
+            PrimDef::FunNpLinalgQr => fun("np_linalg_qr", None),
+            PrimDef::FunNpLinalgSvd => fun("np_linalg_svd", None),
+            PrimDef::FunNpLinalgInv => fun("np_linalg_inv", None),
+            PrimDef::FunNpLinalgPinv => fun("np_linalg_pinv", None),
+            PrimDef::FunNpLinalgMatrixPower => fun("np_linalg_matrix_power", None),
+            PrimDef::FunNpLinalgDet => fun("np_linalg_det", None),
+            PrimDef::FunSpLinalgLu => fun("sp_linalg_lu", None),
+            PrimDef::FunSpLinalgSchur => fun("sp_linalg_schur", None),
+            PrimDef::FunSpLinalgHessenberg => fun("sp_linalg_hessenberg", None),
+
+            // Miscellaneous Python & NAC3 functions
+            PrimDef::FunInt32 => fun("int32", None),
+            PrimDef::FunInt64 => fun("int64", None),
+            PrimDef::FunUInt32 => fun("uint32", None),
+            PrimDef::FunUInt64 => fun("uint64", None),
+            PrimDef::FunFloat => fun("float", None),
+            PrimDef::FunRound => fun("round", None),
+            PrimDef::FunRound64 => fun("round64", None),
+            PrimDef::FunStr => fun("str", None),
+            PrimDef::FunBool => fun("bool", None),
+            PrimDef::FunFloor => fun("floor", None),
+            PrimDef::FunFloor64 => fun("floor64", None),
+            PrimDef::FunCeil => fun("ceil", None),
+            PrimDef::FunCeil64 => fun("ceil64", None),
+            PrimDef::FunLen => fun("len", None),
+            PrimDef::FunMin => fun("min", None),
+            PrimDef::FunMax => fun("max", None),
+            PrimDef::FunAbs => fun("abs", None),
         }
     }
 }
@@ -326,6 +389,9 @@ impl TopLevelDef {
                     r
                 }
             ),
+            TopLevelDef::Variable { name, ty, .. } => {
+                format!("Variable {{ name: {name:?}, ty: {:?} }}", unifier.stringify(*ty),)
+            }
         }
     }
 }
@@ -408,9 +474,9 @@ impl TopLevelComposer {
         let option = unifier.add_ty(TypeEnum::TObj {
             obj_id: PrimDef::Option.id(),
             fields: vec![
-                (PrimDef::OptionIsSome.simple_name().into(), (is_some_type_fun_ty, true)),
-                (PrimDef::OptionIsNone.simple_name().into(), (is_some_type_fun_ty, true)),
-                (PrimDef::OptionUnwrap.simple_name().into(), (unwrap_fun_ty, true)),
+                (PrimDef::FunOptionIsSome.simple_name().into(), (is_some_type_fun_ty, true)),
+                (PrimDef::FunOptionIsNone.simple_name().into(), (is_some_type_fun_ty, true)),
+                (PrimDef::FunOptionUnwrap.simple_name().into(), (unwrap_fun_ty, true)),
             ]
             .into_iter()
             .collect::<HashMap<_, _>>(),
@@ -444,6 +510,7 @@ impl TopLevelComposer {
                 name: "value".into(),
                 ty: ndarray_dtype_tvar.ty,
                 default_value: None,
+                is_vararg: false,
             }],
             ret: none,
             vars: into_var_map([ndarray_dtype_tvar, ndarray_ndims_tvar]),
@@ -451,8 +518,8 @@ impl TopLevelComposer {
         let ndarray = unifier.add_ty(TypeEnum::TObj {
             obj_id: PrimDef::NDArray.id(),
             fields: Mapping::from([
-                (PrimDef::NDArrayCopy.simple_name().into(), (ndarray_copy_fun_ty, true)),
-                (PrimDef::NDArrayFill.simple_name().into(), (ndarray_fill_fun_ty, true)),
+                (PrimDef::FunNDArrayCopy.simple_name().into(), (ndarray_copy_fun_ty, true)),
+                (PrimDef::FunNDArrayFill.simple_name().into(), (ndarray_fill_fun_ty, true)),
             ]),
             params: into_var_map([ndarray_dtype_tvar, ndarray_ndims_tvar]),
         });
@@ -526,6 +593,18 @@ impl TopLevelComposer {
         }
     }
 
+    #[must_use]
+    pub fn make_top_level_variable_def(
+        name: String,
+        simple_name: StrRef,
+        ty: Type,
+        ty_decl: Option<Expr>,
+        resolver: Option<Arc<dyn SymbolResolver + Send + Sync>>,
+        loc: Option<Location>,
+    ) -> TopLevelDef {
+        TopLevelDef::Variable { name, simple_name, ty, ty_decl, resolver, loc }
+    }
+
     #[must_use]
     pub fn make_class_method_name(mut class_name: String, method_name: &str) -> String {
         class_name.push('.');
@@ -671,7 +750,16 @@ impl TopLevelComposer {
         )
     }
 
-    pub fn get_all_assigned_field(stmts: &[Stmt<()>]) -> Result<HashSet<StrRef>, HashSet<String>> {
+    /// This function returns the fields that have been initialized in the `__init__` function of a class
+    /// The function takes as input:
+    ///  * `class_id`: The `object_id` of the class whose function is being evaluated (check `TopLevelDef::Class`)
+    ///  * `definition_ast_list`: A list of ast definitions and statements defined in `TopLevelComposer`
+    ///  * `stmts`: The body of function being parsed. Each statment is analyzed to check varaible initialization statements
+    pub fn get_all_assigned_field(
+        class_id: usize,
+        definition_ast_list: &Vec<DefAst>,
+        stmts: &[Stmt<()>],
+    ) -> Result<HashSet<StrRef>, HashSet<String>> {
         let mut result = HashSet::new();
         for s in stmts {
             match &s.node {
@@ -707,30 +795,138 @@ impl TopLevelComposer {
                 // TODO: do not check for For and While?
                 ast::StmtKind::For { body, orelse, .. }
                 | ast::StmtKind::While { body, orelse, .. } => {
-                    result.extend(Self::get_all_assigned_field(body.as_slice())?);
-                    result.extend(Self::get_all_assigned_field(orelse.as_slice())?);
+                    result.extend(Self::get_all_assigned_field(
+                        class_id,
+                        definition_ast_list,
+                        body.as_slice(),
+                    )?);
+                    result.extend(Self::get_all_assigned_field(
+                        class_id,
+                        definition_ast_list,
+                        orelse.as_slice(),
+                    )?);
                 }
                 ast::StmtKind::If { body, orelse, .. } => {
-                    let inited_for_sure = Self::get_all_assigned_field(body.as_slice())?
-                        .intersection(&Self::get_all_assigned_field(orelse.as_slice())?)
+                    let inited_for_sure = Self::get_all_assigned_field(
+                        class_id,
+                        definition_ast_list,
+                        body.as_slice(),
+                    )?
+                    .intersection(&Self::get_all_assigned_field(
+                        class_id,
+                        definition_ast_list,
+                        orelse.as_slice(),
+                    )?)
                     .copied()
                     .collect::<HashSet<_>>();
                     result.extend(inited_for_sure);
                 }
                 ast::StmtKind::Try { body, orelse, finalbody, .. } => {
-                    let inited_for_sure = Self::get_all_assigned_field(body.as_slice())?
-                        .intersection(&Self::get_all_assigned_field(orelse.as_slice())?)
+                    let inited_for_sure = Self::get_all_assigned_field(
+                        class_id,
+                        definition_ast_list,
+                        body.as_slice(),
+                    )?
+                    .intersection(&Self::get_all_assigned_field(
+                        class_id,
+                        definition_ast_list,
+                        orelse.as_slice(),
+                    )?)
                     .copied()
                     .collect::<HashSet<_>>();
                     result.extend(inited_for_sure);
-                    result.extend(Self::get_all_assigned_field(finalbody.as_slice())?);
+                    result.extend(Self::get_all_assigned_field(
+                        class_id,
+                        definition_ast_list,
+                        finalbody.as_slice(),
+                    )?);
                 }
                 ast::StmtKind::With { body, .. } => {
-                    result.extend(Self::get_all_assigned_field(body.as_slice())?);
+                    result.extend(Self::get_all_assigned_field(
+                        class_id,
+                        definition_ast_list,
+                        body.as_slice(),
+                    )?);
+                }
+                // Variables Initialized in function calls
+                ast::StmtKind::Expr { value, .. } => {
+                    let ExprKind::Call { func, .. } = &value.node else {
+                        continue;
+                    };
+                    let ExprKind::Attribute { value, attr, .. } = &func.node else {
+                        continue;
+                    };
+                    let ExprKind::Name { id, .. } = &value.node else {
+                        continue;
+                    };
+                    // Need to consider the two cases:
+                    //  Case 1) Call to class function i.e. id = `self`
+                    //  Case 2) Call to class ancestor function i.e. id = ancestor_name
+                    // We leave checking whether function in case 2 belonged to class ancestor or not to type checker
+                    //
+                    // According to current handling of `self`, function definition are fixed and do not change regardless
+                    // of which object is passed as `self` i.e. virtual polymorphism is not supported
+                    // Therefore, we change class id for case 2 to reflect behavior of our compiler
+
+                    let class_name = if *id == "self".into() {
+                        let ast::StmtKind::ClassDef { name, .. } =
+                            &definition_ast_list[class_id].1.as_ref().unwrap().node
+                        else {
+                            unreachable!()
+                        };
+                        name
+                    } else {
+                        id
+                    };
+
+                    let parent_method = definition_ast_list.iter().find_map(|def| {
+                        let (
+                            class_def,
+                            Some(ast::Located {
+                                node: ast::StmtKind::ClassDef { name, body, .. },
+                                ..
+                            }),
+                        ) = &def
+                        else {
+                            return None;
+                        };
+                        let TopLevelDef::Class { object_id: class_id, .. } = &*class_def.read()
+                        else {
+                            unreachable!()
+                        };
+
+                        if name == class_name {
+                            body.iter().find_map(|m| {
+                                let ast::StmtKind::FunctionDef { name, body, .. } = &m.node else {
+                                    return None;
+                                };
+                                if *name == *attr {
+                                    return Some((body.clone(), class_id.0));
+                                }
+                                None
+                            })
+                        } else {
+                            None
+                        }
+                    });
+
+                    // If method body is none then method does not exist
+                    if let Some((method_body, class_id)) = parent_method {
+                        result.extend(Self::get_all_assigned_field(
+                            class_id,
+                            definition_ast_list,
+                            method_body.as_slice(),
+                        )?);
+                    } else {
+                        return Err(HashSet::from([format!(
+                            "{}.{} not found in class {class_name} at {}",
+                            *id, *attr, value.location
+                        )]));
+                    }
                 }
                 ast::StmtKind::Pass { .. }
                 | ast::StmtKind::Assert { .. }
-                | ast::StmtKind::Expr { .. } => {}
+                | ast::StmtKind::AnnAssign { .. } => {}
 
                 _ => {
                     unimplemented!()
@@ -938,3 +1134,23 @@ pub fn arraylike_get_ndims(unifier: &mut Unifier, ty: Type) -> u64 {
         _ => 0,
     }
 }
+
+/// Extract an ndarray's `ndims` [type][`Type`] in `u64`. Panic if not possible.
+/// The `ndims` must only contain 1 value.
+#[must_use]
+pub fn extract_ndims(unifier: &Unifier, ndims_ty: Type) -> u64 {
+    let ndims_ty_enum = unifier.get_ty_immutable(ndims_ty);
+    let TypeEnum::TLiteral { values, .. } = &*ndims_ty_enum else {
+        panic!("ndims_ty should be a TLiteral");
+    };
+
+    assert_eq!(values.len(), 1, "ndims_ty TLiteral should only contain 1 value");
+
+    let ndims = values[0].clone();
+    u64::try_from(ndims).unwrap()
+}
+
+/// Return an ndarray's `ndims` as a typechecker [`Type`] from its `u64` value.
+pub fn create_ndims(unifier: &mut Unifier, ndims: u64) -> Type {
+    unifier.get_fresh_literal(vec![SymbolValue::U64(ndims)], None)
+}

nac3core/src/toplevel/mod.rs

@@ -6,36 +6,36 @@ use std::{
     sync::Arc,
 };
 
-use super::codegen::CodeGenContext;
-use super::typecheck::type_inferencer::PrimitiveStore;
-use super::typecheck::typedef::{
-    FunSignature, FuncArg, SharedUnifier, Type, TypeEnum, Unifier, VarMap,
-};
-use crate::{
-    codegen::CodeGenerator,
-    symbol_resolver::{SymbolResolver, ValueEnum},
-    typecheck::{
-        type_inferencer::CodeLocation,
-        typedef::{CallId, TypeVarId},
-    },
-};
 use inkwell::values::BasicValueEnum;
 use itertools::Itertools;
-use nac3parser::ast::{self, Location, Stmt, StrRef};
 use parking_lot::RwLock;
 
-#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Copy, Hash, Debug)]
-pub struct DefinitionId(pub usize);
+use nac3parser::ast::{self, Expr, Location, Stmt, StrRef};
+
+use crate::{
+    codegen::{CodeGenContext, CodeGenerator},
+    symbol_resolver::{SymbolResolver, ValueEnum},
+    typecheck::{
+        type_inferencer::{CodeLocation, PrimitiveStore},
+        typedef::{
+            CallId, FunSignature, FuncArg, SharedUnifier, Type, TypeEnum, TypeVarId, Unifier,
+            VarMap,
+        },
+    },
+};
+use composer::*;
+use type_annotation::*;
 
 pub mod builtins;
 pub mod composer;
 pub mod helper;
 pub mod numpy;
-pub mod type_annotation;
-use composer::*;
-use type_annotation::*;
 #[cfg(test)]
 mod test;
+pub mod type_annotation;
+
+#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Copy, Hash, Debug)]
+pub struct DefinitionId(pub usize);
 
 type GenCallCallback = dyn for<'ctx, 'a> Fn(
         &mut CodeGenContext<'ctx, 'a>,
@@ -148,6 +148,25 @@ pub enum TopLevelDef {
         /// Definition location.
         loc: Option<Location>,
     },
+    Variable {
+        /// Qualified name of the global variable, should be unique globally.
+        name: String,
+
+        /// Simple name, the same as in method/function definition.
+        simple_name: StrRef,
+
+        /// Type of the global variable.
+        ty: Type,
+
+        /// The declared type of the global variable, or [`None`] if no type annotation is provided.
+        ty_decl: Option<Expr>,
+
+        /// Symbol resolver of the module defined the class.
+        resolver: Option<Arc<dyn SymbolResolver + Send + Sync>>,
+
+        /// Definition location.
+        loc: Option<Location>,
+    },
 }
 
 pub struct TopLevelContext {

nac3core/src/toplevel/numpy.rs

@@ -1,11 +1,10 @@
-use crate::{
-    toplevel::helper::PrimDef,
-    typecheck::{
+use itertools::Itertools;
+
+use super::helper::PrimDef;
+use crate::typecheck::{
     type_inferencer::PrimitiveStore,
     typedef::{Type, TypeEnum, TypeVarId, Unifier, VarMap},
-    },
 };
-use itertools::Itertools;
 
 /// Creates a `ndarray` [`Type`] with the given type arguments.
 ///

nac3core/src/toplevel/snapshots/nac3core__toplevel__test__test_analyze__generic_class.snap

@@ -5,7 +5,7 @@ expression: res_vec
 [
     "Class {\nname: \"Generic_A\",\nancestors: [\"Generic_A[V]\", \"B\"],\nfields: [\"aa\", \"a\"],\nmethods: [(\"__init__\", \"fn[[], none]\"), (\"foo\", \"fn[[b:T], none]\"), (\"fun\", \"fn[[a:int32], V]\")],\ntype_vars: [\"V\"]\n}\n",
     "Function {\nname: \"Generic_A.__init__\",\nsig: \"fn[[], none]\",\nvar_id: []\n}\n",
-    "Function {\nname: \"Generic_A.fun\",\nsig: \"fn[[a:int32], V]\",\nvar_id: [TypeVarId(245)]\n}\n",
+    "Function {\nname: \"Generic_A.fun\",\nsig: \"fn[[a:int32], V]\",\nvar_id: [TypeVarId(241)]\n}\n",
     "Class {\nname: \"B\",\nancestors: [\"B\"],\nfields: [\"aa\"],\nmethods: [(\"__init__\", \"fn[[], none]\"), (\"foo\", \"fn[[b:T], none]\")],\ntype_vars: []\n}\n",
     "Function {\nname: \"B.__init__\",\nsig: \"fn[[], none]\",\nvar_id: []\n}\n",
     "Function {\nname: \"B.foo\",\nsig: \"fn[[b:T], none]\",\nvar_id: []\n}\n",

nac3core/src/toplevel/snapshots/nac3core__toplevel__test__test_analyze__inheritance_override.snap

@@ -7,7 +7,7 @@ expression: res_vec
     "Function {\nname: \"A.__init__\",\nsig: \"fn[[t:T], none]\",\nvar_id: []\n}\n",
     "Function {\nname: \"A.fun\",\nsig: \"fn[[a:int32, b:T], list[virtual[B[bool]]]]\",\nvar_id: []\n}\n",
     "Function {\nname: \"A.foo\",\nsig: \"fn[[c:C], none]\",\nvar_id: []\n}\n",
-    "Class {\nname: \"B\",\nancestors: [\"B[typevar234]\", \"A[float]\"],\nfields: [\"a\", \"b\", \"c\", \"d\"],\nmethods: [(\"__init__\", \"fn[[], none]\"), (\"fun\", \"fn[[a:int32, b:T], list[virtual[B[bool]]]]\"), (\"foo\", \"fn[[c:C], none]\")],\ntype_vars: [\"typevar234\"]\n}\n",
+    "Class {\nname: \"B\",\nancestors: [\"B[typevar230]\", \"A[float]\"],\nfields: [\"a\", \"b\", \"c\", \"d\"],\nmethods: [(\"__init__\", \"fn[[], none]\"), (\"fun\", \"fn[[a:int32, b:T], list[virtual[B[bool]]]]\"), (\"foo\", \"fn[[c:C], none]\")],\ntype_vars: [\"typevar230\"]\n}\n",
     "Function {\nname: \"B.__init__\",\nsig: \"fn[[], none]\",\nvar_id: []\n}\n",
     "Function {\nname: \"B.fun\",\nsig: \"fn[[a:int32, b:T], list[virtual[B[bool]]]]\",\nvar_id: []\n}\n",
     "Class {\nname: \"C\",\nancestors: [\"C\", \"B[bool]\", \"A[float]\"],\nfields: [\"a\", \"b\", \"c\", \"d\", \"e\"],\nmethods: [(\"__init__\", \"fn[[], none]\"), (\"fun\", \"fn[[a:int32, b:T], list[virtual[B[bool]]]]\"), (\"foo\", \"fn[[c:C], none]\")],\ntype_vars: []\n}\n",

nac3core/src/toplevel/snapshots/nac3core__toplevel__test__test_analyze__list_tuple_generic.snap

@@ -5,8 +5,8 @@ expression: res_vec
 [
     "Function {\nname: \"foo\",\nsig: \"fn[[a:list[int32], b:tuple[T, float]], A[B, bool]]\",\nvar_id: []\n}\n",
     "Class {\nname: \"A\",\nancestors: [\"A[T, V]\"],\nfields: [\"a\", \"b\"],\nmethods: [(\"__init__\", \"fn[[v:V], none]\"), (\"fun\", \"fn[[a:T], V]\")],\ntype_vars: [\"T\", \"V\"]\n}\n",
-    "Function {\nname: \"A.__init__\",\nsig: \"fn[[v:V], none]\",\nvar_id: [TypeVarId(247)]\n}\n",
-    "Function {\nname: \"A.fun\",\nsig: \"fn[[a:T], V]\",\nvar_id: [TypeVarId(252)]\n}\n",
+    "Function {\nname: \"A.__init__\",\nsig: \"fn[[v:V], none]\",\nvar_id: [TypeVarId(243)]\n}\n",
+    "Function {\nname: \"A.fun\",\nsig: \"fn[[a:T], V]\",\nvar_id: [TypeVarId(248)]\n}\n",
     "Function {\nname: \"gfun\",\nsig: \"fn[[a:A[list[float], int32]], none]\",\nvar_id: []\n}\n",
     "Class {\nname: \"B\",\nancestors: [\"B\"],\nfields: [],\nmethods: [(\"__init__\", \"fn[[], none]\")],\ntype_vars: []\n}\n",
     "Function {\nname: \"B.__init__\",\nsig: \"fn[[], none]\",\nvar_id: []\n}\n",

nac3core/src/toplevel/snapshots/nac3core__toplevel__test__test_analyze__self1.snap

@@ -3,7 +3,7 @@ source: nac3core/src/toplevel/test.rs
 expression: res_vec
 ---
 [
-    "Class {\nname: \"A\",\nancestors: [\"A[typevar233, typevar234]\"],\nfields: [\"a\", \"b\"],\nmethods: [(\"__init__\", \"fn[[a:A[float, bool], b:B], none]\"), (\"fun\", \"fn[[a:A[float, bool]], A[bool, int32]]\")],\ntype_vars: [\"typevar233\", \"typevar234\"]\n}\n",
+    "Class {\nname: \"A\",\nancestors: [\"A[typevar229, typevar230]\"],\nfields: [\"a\", \"b\"],\nmethods: [(\"__init__\", \"fn[[a:A[float, bool], b:B], none]\"), (\"fun\", \"fn[[a:A[float, bool]], A[bool, int32]]\")],\ntype_vars: [\"typevar229\", \"typevar230\"]\n}\n",
     "Function {\nname: \"A.__init__\",\nsig: \"fn[[a:A[float, bool], b:B], none]\",\nvar_id: []\n}\n",
     "Function {\nname: \"A.fun\",\nsig: \"fn[[a:A[float, bool]], A[bool, int32]]\",\nvar_id: []\n}\n",
     "Class {\nname: \"B\",\nancestors: [\"B\", \"A[int64, bool]\"],\nfields: [\"a\", \"b\"],\nmethods: [(\"__init__\", \"fn[[], none]\"), (\"fun\", \"fn[[a:A[float, bool]], A[bool, int32]]\"), (\"foo\", \"fn[[b:B], B]\"), (\"bar\", \"fn[[a:A[list[B], int32]], tuple[A[virtual[A[B, int32]], bool], B]]\")],\ntype_vars: []\n}\n",

nac3core/src/toplevel/snapshots/nac3core__toplevel__test__test_analyze__simple_class_compose.snap

@@ -6,12 +6,12 @@ expression: res_vec
     "Class {\nname: \"A\",\nancestors: [\"A\"],\nfields: [\"a\"],\nmethods: [(\"__init__\", \"fn[[], none]\"), (\"fun\", \"fn[[b:B], none]\"), (\"foo\", \"fn[[a:T, b:V], none]\")],\ntype_vars: []\n}\n",
     "Function {\nname: \"A.__init__\",\nsig: \"fn[[], none]\",\nvar_id: []\n}\n",
     "Function {\nname: \"A.fun\",\nsig: \"fn[[b:B], none]\",\nvar_id: []\n}\n",
-    "Function {\nname: \"A.foo\",\nsig: \"fn[[a:T, b:V], none]\",\nvar_id: [TypeVarId(253)]\n}\n",
+    "Function {\nname: \"A.foo\",\nsig: \"fn[[a:T, b:V], none]\",\nvar_id: [TypeVarId(249)]\n}\n",
     "Class {\nname: \"B\",\nancestors: [\"B\", \"C\", \"A\"],\nfields: [\"a\"],\nmethods: [(\"__init__\", \"fn[[], none]\"), (\"fun\", \"fn[[b:B], none]\"), (\"foo\", \"fn[[a:T, b:V], none]\")],\ntype_vars: []\n}\n",
     "Function {\nname: \"B.__init__\",\nsig: \"fn[[], none]\",\nvar_id: []\n}\n",
     "Class {\nname: \"C\",\nancestors: [\"C\", \"A\"],\nfields: [\"a\"],\nmethods: [(\"__init__\", \"fn[[], none]\"), (\"fun\", \"fn[[b:B], none]\"), (\"foo\", \"fn[[a:T, b:V], none]\")],\ntype_vars: []\n}\n",
     "Function {\nname: \"C.__init__\",\nsig: \"fn[[], none]\",\nvar_id: []\n}\n",
     "Function {\nname: \"C.fun\",\nsig: \"fn[[b:B], none]\",\nvar_id: []\n}\n",
     "Function {\nname: \"foo\",\nsig: \"fn[[a:A], none]\",\nvar_id: []\n}\n",
-    "Function {\nname: \"ff\",\nsig: \"fn[[a:T], V]\",\nvar_id: [TypeVarId(261)]\n}\n",
+    "Function {\nname: \"ff\",\nsig: \"fn[[a:T], V]\",\nvar_id: [TypeVarId(257)]\n}\n",
 ]

nac3core/src/toplevel/test.rs

@@ -1,21 +1,23 @@
-use super::*;
-use crate::toplevel::helper::PrimDef;
-use crate::typecheck::typedef::into_var_map;
+use std::{collections::HashMap, sync::Arc};
+
+use indoc::indoc;
+use parking_lot::Mutex;
+use test_case::test_case;
+
+use nac3parser::{
+    ast::{fold::Fold, FileName},
+    parser::parse_program,
+};
+
+use super::{helper::PrimDef, DefinitionId, *};
 use crate::{
     codegen::CodeGenContext,
     symbol_resolver::{SymbolResolver, ValueEnum},
-    toplevel::DefinitionId,
     typecheck::{
         type_inferencer::PrimitiveStore,
-        typedef::{Type, Unifier},
+        typedef::{into_var_map, Type, Unifier},
     },
 };
-use indoc::indoc;
-use nac3parser::ast::FileName;
-use nac3parser::{ast::fold::Fold, parser::parse_program};
-use parking_lot::Mutex;
-use std::{collections::HashMap, sync::Arc};
-use test_case::test_case;
 
 struct ResolverInternal {
     id_to_type: Mutex<HashMap<StrRef, Type>>,
@@ -62,6 +64,7 @@ impl SymbolResolver for Resolver {
         &self,
         _: StrRef,
         _: &mut CodeGenContext<'ctx, '_>,
+        _: &mut dyn CodeGenerator,
     ) -> Option<ValueEnum<'ctx>> {
         unimplemented!()
     }
@@ -117,7 +120,8 @@ impl SymbolResolver for Resolver {
     "register"
 )]
 fn test_simple_register(source: Vec<&str>) {
-    let mut composer = TopLevelComposer::new(Vec::new(), ComposerConfig::default(), 64).0;
+    let mut composer =
+        TopLevelComposer::new(Vec::new(), Vec::new(), ComposerConfig::default(), 64).0;
 
     for s in source {
         let ast = parse_program(s, FileName::default()).unwrap();
@@ -137,7 +141,8 @@ fn test_simple_register(source: Vec<&str>) {
     "register"
 )]
 fn test_simple_register_without_constructor(source: &str) {
-    let mut composer = TopLevelComposer::new(Vec::new(), ComposerConfig::default(), 64).0;
+    let mut composer =
+        TopLevelComposer::new(Vec::new(), Vec::new(), ComposerConfig::default(), 64).0;
     let ast = parse_program(source, FileName::default()).unwrap();
     let ast = ast[0].clone();
     composer.register_top_level(ast, None, "", true).unwrap();
@@ -171,7 +176,8 @@ fn test_simple_register_without_constructor(source: &str) {
     "function compose"
 )]
 fn test_simple_function_analyze(source: &[&str], tys: &[&str], names: &[&str]) {
-    let mut composer = TopLevelComposer::new(Vec::new(), ComposerConfig::default(), 64).0;
+    let mut composer =
+        TopLevelComposer::new(Vec::new(), Vec::new(), ComposerConfig::default(), 64).0;
 
     let internal_resolver = Arc::new(ResolverInternal {
         id_to_def: Mutex::default(),
@@ -519,7 +525,8 @@ fn test_simple_function_analyze(source: &[&str], tys: &[&str], names: &[&str]) {
 )]
 fn test_analyze(source: &[&str], res: &[&str]) {
     let print = false;
-    let mut composer = TopLevelComposer::new(Vec::new(), ComposerConfig::default(), 64).0;
+    let mut composer =
+        TopLevelComposer::new(Vec::new(), Vec::new(), ComposerConfig::default(), 64).0;
 
     let internal_resolver = make_internal_resolver_with_tvar(
         vec![
@@ -696,7 +703,8 @@ fn test_analyze(source: &[&str], res: &[&str]) {
 )]
 fn test_inference(source: Vec<&str>, res: &[&str]) {
     let print = true;
-    let mut composer = TopLevelComposer::new(Vec::new(), ComposerConfig::default(), 64).0;
+    let mut composer =
+        TopLevelComposer::new(Vec::new(), Vec::new(), ComposerConfig::default(), 64).0;
 
     let internal_resolver = make_internal_resolver_with_tvar(
         vec![

nac3core/src/toplevel/type_annotation.rs

@@ -1,9 +1,13 @@
-use super::*;
-use crate::symbol_resolver::SymbolValue;
-use crate::toplevel::helper::PrimDef;
-use crate::typecheck::typedef::VarMap;
+use strum::IntoEnumIterator;
+
 use nac3parser::ast::Constant;
 
+use super::{
+    helper::{PrimDef, PrimDefDetails},
+    *,
+};
+use crate::{symbol_resolver::SymbolValue, typecheck::typedef::VarMap};
+
 #[derive(Clone, Debug)]
 pub enum TypeAnnotation {
     Primitive(Type),
@@ -357,6 +361,7 @@ pub fn parse_ast_to_type_annotation_kinds<T, S: std::hash::BuildHasher + Clone>(
 pub fn get_type_from_type_annotation_kinds(
     top_level_defs: &[Arc<RwLock<TopLevelDef>>],
     unifier: &mut Unifier,
+    primitives: &PrimitiveStore,
     ann: &TypeAnnotation,
     subst_list: &mut Option<Vec<Type>>,
 ) -> Result<Type, HashSet<String>> {
@@ -379,10 +384,43 @@ pub fn get_type_from_type_annotation_kinds(
             let param_ty = params
                 .iter()
                 .map(|x| {
-                    get_type_from_type_annotation_kinds(top_level_defs, unifier, x, subst_list)
+                    get_type_from_type_annotation_kinds(
+                        top_level_defs,
+                        unifier,
+                        primitives,
+                        x,
+                        subst_list,
+                    )
                 })
                 .collect::<Result<Vec<_>, _>>()?;
 
+            let ty = if let Some(prim_def) = PrimDef::iter().find(|prim| prim.id() == *obj_id) {
+                // Primitive TopLevelDefs do not contain all fields that are present in their Type
+                // counterparts, so directly perform subst on the Type instead.
+
+                let PrimDefDetails::PrimClass { get_ty_fn, .. } = prim_def.details() else {
+                    unreachable!()
+                };
+
+                let base_ty = get_ty_fn(primitives);
+                let params =
+                    if let TypeEnum::TObj { params, .. } = &*unifier.get_ty_immutable(base_ty) {
+                        params.clone()
+                    } else {
+                        unreachable!()
+                    };
+
+                unifier
+                    .subst(
+                        get_ty_fn(primitives),
+                        &params
+                            .iter()
+                            .zip(param_ty)
+                            .map(|(obj_tv, param)| (*obj_tv.0, param))
+                            .collect(),
+                    )
+                    .unwrap_or(base_ty)
+            } else {
                 let subst = {
                     // check for compatible range
                     // TODO: if allow type var to be applied(now this disallowed in the parse_to_type_annotation), need more check
@@ -424,12 +462,15 @@ pub fn get_type_from_type_annotation_kinds(
                                 }
                             }
 
-                        TypeEnum::TVar { id, range, name, loc, is_const_generic: true, .. } => {
+                            TypeEnum::TVar {
+                                id, range, name, loc, is_const_generic: true, ..
+                            } => {
                                 let ty = range[0];
                                 let ok: bool = {
                                     // create a temp type var and unify to check compatibility
                                     p == *tvar || {
-                                    let temp = unifier.get_fresh_const_generic_var(ty, *name, *loc);
+                                        let temp =
+                                            unifier.get_fresh_const_generic_var(ty, *name, *loc);
                                         unifier.unify(temp.ty, p).is_ok()
                                     }
                                 };
@@ -468,11 +509,16 @@ pub fn get_type_from_type_annotation_kinds(
                     fields: tobj_fields,
                     params: subst,
                 });
+
                 if need_subst {
                     if let Some(wl) = subst_list.as_mut() {
                         wl.push(ty);
                     }
                 }
+
+                ty
+            };
+
             Ok(ty)
         }
         TypeAnnotation::Primitive(ty) | TypeAnnotation::TypeVar(ty) => Ok(*ty),
@@ -490,6 +536,7 @@ pub fn get_type_from_type_annotation_kinds(
             let ty = get_type_from_type_annotation_kinds(
                 top_level_defs,
                 unifier,
+                primitives,
                 ty.as_ref(),
                 subst_list,
             )?;
@@ -499,10 +546,16 @@ pub fn get_type_from_type_annotation_kinds(
             let tys = tys
                 .iter()
                 .map(|x| {
-                    get_type_from_type_annotation_kinds(top_level_defs, unifier, x, subst_list)
+                    get_type_from_type_annotation_kinds(
+                        top_level_defs,
+                        unifier,
+                        primitives,
+                        x,
+                        subst_list,
+                    )
                 })
                 .collect::<Result<Vec<_>, _>>()?;
-            Ok(unifier.add_ty(TypeEnum::TTuple { ty: tys }))
+            Ok(unifier.add_ty(TypeEnum::TTuple { ty: tys, is_vararg_ctx: false }))
         }
     }
 }

nac3core/src/typecheck/function_check.rs

@@ -1,13 +1,19 @@
-use crate::toplevel::helper::PrimDef;
+use std::{
+    collections::{HashMap, HashSet},
+    iter::once,
+};
 
-use super::type_inferencer::Inferencer;
-use super::typedef::{Type, TypeEnum};
 use nac3parser::ast::{
     self, Constant, Expr, ExprKind,
     Operator::{LShift, RShift},
     Stmt, StmtKind, StrRef,
 };
-use std::{collections::HashSet, iter::once};
+
+use super::{
+    type_inferencer::{DeclarationSource, IdentifierInfo, Inferencer},
+    typedef::{Type, TypeEnum},
+};
+use crate::toplevel::helper::PrimDef;
 
 impl<'a> Inferencer<'a> {
     fn should_have_value(&mut self, expr: &Expr<Option<Type>>) -> Result<(), HashSet<String>> {
@@ -21,26 +27,45 @@ impl<'a> Inferencer<'a> {
     fn check_pattern(
         &mut self,
         pattern: &Expr<Option<Type>>,
-        defined_identifiers: &mut HashSet<StrRef>,
+        defined_identifiers: &mut HashMap<StrRef, IdentifierInfo>,
     ) -> Result<(), HashSet<String>> {
         match &pattern.node {
             ExprKind::Name { id, .. } if id == &"none".into() => {
                 Err(HashSet::from([format!("cannot assign to a `none` (at {})", pattern.location)]))
             }
             ExprKind::Name { id, .. } => {
-                if !defined_identifiers.contains(id) {
-                    defined_identifiers.insert(*id);
+                // If `id` refers to a declared symbol, reject this assignment if it is used in the
+                // context of an (implicit) global variable
+                if let Some(id_info) = defined_identifiers.get(id) {
+                    if matches!(
+                        id_info.source,
+                        DeclarationSource::Global { is_explicit: Some(false) }
+                    ) {
+                        return Err(HashSet::from([format!(
+                            "cannot access local variable '{id}' before it is declared (at {})",
+                            pattern.location
+                        )]));
+                    }
+                }
+
+                if !defined_identifiers.contains_key(id) {
+                    defined_identifiers.insert(*id, IdentifierInfo::default());
                 }
                 self.should_have_value(pattern)?;
                 Ok(())
             }
-            ExprKind::Tuple { elts, .. } => {
+            ExprKind::List { elts, .. } | ExprKind::Tuple { elts, .. } => {
                 for elt in elts {
                     self.check_pattern(elt, defined_identifiers)?;
                     self.should_have_value(elt)?;
                 }
                 Ok(())
             }
+            ExprKind::Starred { value, .. } => {
+                self.check_pattern(value, defined_identifiers)?;
+                self.should_have_value(value)?;
+                Ok(())
+            }
             ExprKind::Subscript { value, slice, .. } => {
                 self.check_expr(value, defined_identifiers)?;
                 self.should_have_value(value)?;
@@ -64,7 +89,7 @@ impl<'a> Inferencer<'a> {
     fn check_expr(
         &mut self,
         expr: &Expr<Option<Type>>,
-        defined_identifiers: &mut HashSet<StrRef>,
+        defined_identifiers: &mut HashMap<StrRef, IdentifierInfo>,
     ) -> Result<(), HashSet<String>> {
         // there are some cases where the custom field is None
         if let Some(ty) = &expr.custom {
@@ -85,7 +110,7 @@ impl<'a> Inferencer<'a> {
                     return Ok(());
                 }
                 self.should_have_value(expr)?;
-                if !defined_identifiers.contains(id) {
+                if !defined_identifiers.contains_key(id) {
                     match self.function_data.resolver.get_symbol_type(
                         self.unifier,
                         &self.top_level.definitions.read(),
@@ -93,7 +118,22 @@ impl<'a> Inferencer<'a> {
                         *id,
                     ) {
                         Ok(_) => {
-                            self.defined_identifiers.insert(*id);
+                            let is_global = self.is_id_global(*id);
+
+                            defined_identifiers.insert(
+                                *id,
+                                IdentifierInfo {
+                                    source: match is_global {
+                                        Some(true) => {
+                                            DeclarationSource::Global { is_explicit: Some(false) }
+                                        }
+                                        Some(false) => {
+                                            DeclarationSource::Global { is_explicit: None }
+                                        }
+                                        None => DeclarationSource::Local,
+                                    },
+                                },
+                            );
                         }
                         Err(e) => {
                             return Err(HashSet::from([format!(
@@ -166,9 +206,7 @@ impl<'a> Inferencer<'a> {
                 let mut defined_identifiers = defined_identifiers.clone();
                 for arg in &args.args {
                     // TODO: should we check the types here?
-                    if !defined_identifiers.contains(&arg.node.arg) {
-                        defined_identifiers.insert(arg.node.arg);
-                    }
+                    defined_identifiers.entry(arg.node.arg).or_default();
                 }
                 self.check_expr(body, &mut defined_identifiers)?;
             }
@@ -207,6 +245,9 @@ impl<'a> Inferencer<'a> {
     /// This is a workaround preventing the caller from using a variable `alloca`-ed in the body, which
     /// is freed when the function returns.
     fn check_return_value_ty(&mut self, ret_ty: Type) -> bool {
+        if cfg!(feature = "no-escape-analysis") {
+            true
+        } else {
             match &*self.unifier.get_ty_immutable(ret_ty) {
                 TypeEnum::TObj { .. } => [
                     self.primitives.int32,
@@ -218,16 +259,17 @@ impl<'a> Inferencer<'a> {
                 ]
                 .iter()
                 .any(|allowed_ty| self.unifier.unioned(ret_ty, *allowed_ty)),
-            TypeEnum::TTuple { ty } => ty.iter().all(|t| self.check_return_value_ty(*t)),
+                TypeEnum::TTuple { ty, .. } => ty.iter().all(|t| self.check_return_value_ty(*t)),
                 _ => false,
             }
         }
+    }
 
     // check statements for proper identifier def-use and return on all paths
     fn check_stmt(
         &mut self,
         stmt: &Stmt<Option<Type>>,
-        defined_identifiers: &mut HashSet<StrRef>,
+        defined_identifiers: &mut HashMap<StrRef, IdentifierInfo>,
     ) -> Result<bool, HashSet<String>> {
         match &stmt.node {
             StmtKind::For { target, iter, body, orelse, .. } => {
@@ -253,9 +295,11 @@ impl<'a> Inferencer<'a> {
                 let body_returned = self.check_block(body, &mut body_identifiers)?;
                 let orelse_returned = self.check_block(orelse, &mut orelse_identifiers)?;
 
-                for ident in &body_identifiers {
-                    if !defined_identifiers.contains(ident) && orelse_identifiers.contains(ident) {
-                        defined_identifiers.insert(*ident);
+                for ident in body_identifiers.keys() {
+                    if !defined_identifiers.contains_key(ident)
+                        && orelse_identifiers.contains_key(ident)
+                    {
+                        defined_identifiers.insert(*ident, IdentifierInfo::default());
                     }
                 }
                 Ok(body_returned && orelse_returned)
@@ -286,7 +330,7 @@ impl<'a> Inferencer<'a> {
                     let mut defined_identifiers = defined_identifiers.clone();
                     let ast::ExcepthandlerKind::ExceptHandler { name, body, .. } = &handler.node;
                     if let Some(name) = name {
-                        defined_identifiers.insert(*name);
+                        defined_identifiers.insert(*name, IdentifierInfo::default());
                     }
                     self.check_block(body, &mut defined_identifiers)?;
                 }
@@ -350,6 +394,44 @@ impl<'a> Inferencer<'a> {
                 }
                 Ok(true)
             }
+            StmtKind::Global { names, .. } => {
+                for id in names {
+                    if let Some(id_info) = defined_identifiers.get(id) {
+                        if id_info.source == DeclarationSource::Local {
+                            return Err(HashSet::from([format!(
+                                "name '{id}' is referenced prior to global declaration at {}",
+                                stmt.location,
+                            )]));
+                        }
+
+                        continue;
+                    }
+
+                    match self.function_data.resolver.get_symbol_type(
+                        self.unifier,
+                        &self.top_level.definitions.read(),
+                        self.primitives,
+                        *id,
+                    ) {
+                        Ok(_) => {
+                            defined_identifiers.insert(
+                                *id,
+                                IdentifierInfo {
+                                    source: DeclarationSource::Global { is_explicit: Some(true) },
+                                },
+                            );
+                        }
+                        Err(e) => {
+                            return Err(HashSet::from([format!(
+                                "type error at identifier `{}` ({}) at {}",
+                                id, e, stmt.location
+                            )]))
+                        }
+                    }
+                }
+
+                Ok(false)
+            }
             // break, raise, etc.
             _ => Ok(false),
         }
@@ -358,7 +440,7 @@ impl<'a> Inferencer<'a> {
     pub fn check_block(
         &mut self,
         block: &[Stmt<Option<Type>>],
-        defined_identifiers: &mut HashSet<StrRef>,
+        defined_identifiers: &mut HashMap<StrRef, IdentifierInfo>,
     ) -> Result<bool, HashSet<String>> {
         let mut ret = false;
         for stmt in block {

nac3core/src/typecheck/magic_methods.rs

@@ -1,17 +1,21 @@
-use crate::symbol_resolver::SymbolValue;
-use crate::toplevel::helper::PrimDef;
-use crate::toplevel::numpy::{make_ndarray_ty, unpack_ndarray_var_tys};
-use crate::typecheck::{
+use std::{cmp::max, collections::HashMap, rc::Rc};
+
+use itertools::{iproduct, Itertools};
+use strum::IntoEnumIterator;
+
+use nac3parser::ast::{Cmpop, Operator, StrRef, Unaryop};
+
+use super::{
     type_inferencer::*,
     typedef::{FunSignature, FuncArg, Type, TypeEnum, Unifier, VarMap},
 };
-use itertools::{iproduct, Itertools};
-use nac3parser::ast::StrRef;
-use nac3parser::ast::{Cmpop, Operator, Unaryop};
-use std::cmp::max;
-use std::collections::HashMap;
-use std::rc::Rc;
-use strum::IntoEnumIterator;
+use crate::{
+    symbol_resolver::SymbolValue,
+    toplevel::{
+        helper::PrimDef,
+        numpy::{make_ndarray_ty, unpack_ndarray_var_tys},
+    },
+};
 
 /// The variant of a binary operator.
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
@@ -197,6 +201,7 @@ pub fn impl_binop(
                             ty: other_ty,
                             default_value: None,
                             name: "other".into(),
+                            is_vararg: false,
                         }],
                     })),
                     false,
@@ -261,6 +266,7 @@ pub fn impl_cmpop(
                             ty: other_ty,
                             default_value: None,
                             name: "other".into(),
+                            is_vararg: false,
                         }],
                     })),
                     false,
@@ -518,6 +524,23 @@ pub fn typeof_binop(
         }
 
         Operator::MatMult => {
+            // NOTE: NumPy matmul's LHS and RHS must both be ndarrays. Scalars are not allowed.
+            match (&*unifier.get_ty(lhs), &*unifier.get_ty(rhs)) {
+                (
+                    TypeEnum::TObj { obj_id: lhs_obj_id, .. },
+                    TypeEnum::TObj { obj_id: rhs_obj_id, .. },
+                ) if *lhs_obj_id == primitives.ndarray.obj_id(unifier).unwrap()
+                    && *rhs_obj_id == primitives.ndarray.obj_id(unifier).unwrap() =>
+                {
+                    // LHS and RHS have valid types
+                }
+                _ => {
+                    let lhs_str = unifier.stringify(lhs);
+                    let rhs_str = unifier.stringify(rhs);
+                    return Err(format!("ndarray.__matmul__ only accepts ndarray operands, but left operand has type {lhs_str}, and right operand has type {rhs_str}"));
+                }
+            }
+
             let (_, lhs_ndims) = unpack_ndarray_var_tys(unifier, lhs);
             let lhs_ndims = match &*unifier.get_ty_immutable(lhs_ndims) {
                 TypeEnum::TLiteral { values, .. } => {
@@ -678,6 +701,7 @@ pub fn set_primitives_magic_methods(store: &PrimitiveStore, unifier: &mut Unifie
         bool: bool_t,
         uint32: uint32_t,
         uint64: uint64_t,
+        str: str_t,
         list: list_t,
         ndarray: ndarray_t,
         ..
@@ -723,6 +747,9 @@ pub fn set_primitives_magic_methods(store: &PrimitiveStore, unifier: &mut Unifie
     impl_sign(unifier, store, bool_t, Some(int32_t));
     impl_eq(unifier, store, bool_t, &[bool_t, ndarray_bool_t], None);
 
+    /* str ========= */
+    impl_cmpop(unifier, store, str_t, &[str_t], &[Cmpop::Eq, Cmpop::NotEq], Some(bool_t));
+
     /* list ======== */
     impl_binop(unifier, store, list_t, &[list_t], Some(list_t), &[Operator::Add]);
     impl_binop(unifier, store, list_t, &[int32_t, int64_t], Some(list_t), &[Operator::Mult]);

nac3core/src/typecheck/type_error.rs

@@ -1,14 +1,13 @@
-use std::collections::HashMap;
-use std::fmt::Display;
+use std::{collections::HashMap, fmt::Display};
 
-use crate::typecheck::{magic_methods::HasOpInfo, typedef::TypeEnum};
+use itertools::Itertools;
+
+use nac3parser::ast::{Cmpop, Location, StrRef};
 
 use super::{
-    magic_methods::Binop,
-    typedef::{RecordKey, Type, Unifier},
+    magic_methods::{Binop, HasOpInfo},
+    typedef::{RecordKey, Type, TypeEnum, Unifier},
 };
-use itertools::Itertools;
-use nac3parser::ast::{Cmpop, Location, StrRef};
 
 #[derive(Debug, Clone)]
 pub enum TypeErrorKind {
@@ -183,9 +182,10 @@ impl<'a> Display for DisplayTypeError<'a> {
                         }
                         result
                     }
-                    (TypeEnum::TTuple { ty: ty1 }, TypeEnum::TTuple { ty: ty2 })
-                        if ty1.len() != ty2.len() =>
-                    {
+                    (
+                        TypeEnum::TTuple { ty: ty1, is_vararg_ctx: is_vararg1 },
+                        TypeEnum::TTuple { ty: ty2, is_vararg_ctx: is_vararg2 },
+                    ) if !is_vararg1 && !is_vararg2 && ty1.len() != ty2.len() => {
                         let t1 = self.unifier.stringify_with_notes(*t1, &mut notes);
                         let t2 = self.unifier.stringify_with_notes(*t2, &mut notes);
                         write!(f, "Tuple length mismatch: got {t1} and {t2}")

nac3core/src/typecheck/type_inferencer/test.rs

@@ -1,18 +1,20 @@
-use super::super::{magic_methods::with_fields, typedef::*};
-use super::*;
-use crate::{
-    codegen::CodeGenContext,
-    symbol_resolver::ValueEnum,
-    toplevel::{helper::PrimDef, DefinitionId, TopLevelDef},
-};
+use std::iter::zip;
+
 use indexmap::IndexMap;
 use indoc::indoc;
-use nac3parser::ast::FileName;
-use nac3parser::parser::parse_program;
 use parking_lot::RwLock;
-use std::iter::zip;
 use test_case::test_case;
 
+use nac3parser::{ast::FileName, parser::parse_program};
+
+use super::*;
+use crate::{
+    codegen::{CodeGenContext, CodeGenerator},
+    symbol_resolver::ValueEnum,
+    toplevel::{helper::PrimDef, DefinitionId, TopLevelDef},
+    typecheck::{magic_methods::with_fields, typedef::*},
+};
+
 struct Resolver {
     id_to_type: HashMap<StrRef, Type>,
     id_to_def: HashMap<StrRef, DefinitionId>,
@@ -41,6 +43,7 @@ impl SymbolResolver for Resolver {
         &self,
         _: StrRef,
         _: &mut CodeGenContext<'ctx, '_>,
+        _: &mut dyn CodeGenerator,
     ) -> Option<ValueEnum<'ctx>> {
         unimplemented!()
     }
@@ -83,7 +86,12 @@ impl TestEnvironment {
         });
         with_fields(&mut unifier, int32, |unifier, fields| {
             let add_ty = unifier.add_ty(TypeEnum::TFunc(FunSignature {
-                args: vec![FuncArg { name: "other".into(), ty: int32, default_value: None }],
+                args: vec![FuncArg {
+                    name: "other".into(),
+                    ty: int32,
+                    default_value: None,
+                    is_vararg: false,
+                }],
                 ret: int32,
                 vars: VarMap::new(),
             }));
@@ -224,7 +232,12 @@ impl TestEnvironment {
         });
         with_fields(&mut unifier, int32, |unifier, fields| {
             let add_ty = unifier.add_ty(TypeEnum::TFunc(FunSignature {
-                args: vec![FuncArg { name: "other".into(), ty: int32, default_value: None }],
+                args: vec![FuncArg {
+                    name: "other".into(),
+                    ty: int32,
+                    default_value: None,
+                    is_vararg: false,
+                }],
                 ret: int32,
                 vars: VarMap::new(),
             }));
@@ -507,7 +520,7 @@ impl TestEnvironment {
             primitives: &mut self.primitives,
             virtual_checks: &mut self.virtual_checks,
             calls: &mut self.calls,
-            defined_identifiers: HashSet::default(),
+            defined_identifiers: HashMap::default(),
             in_handler: false,
         }
     }
@@ -583,8 +596,9 @@ fn test_basic(source: &str, mapping: &HashMap<&str, &str>, virtuals: &[(&str, &s
     println!("source:\n{source}");
     let mut env = TestEnvironment::new();
     let id_to_name = std::mem::take(&mut env.id_to_name);
-    let mut defined_identifiers: HashSet<_> = env.identifier_mapping.keys().copied().collect();
-    defined_identifiers.insert("virtual".into());
+    let mut defined_identifiers: HashMap<_, _> =
+        env.identifier_mapping.keys().copied().map(|id| (id, IdentifierInfo::default())).collect();
+    defined_identifiers.insert("virtual".into(), IdentifierInfo::default());
     let mut inferencer = env.get_inferencer();
     inferencer.defined_identifiers.clone_from(&defined_identifiers);
     let statements = parse_program(source, FileName::default()).unwrap();
@@ -729,8 +743,9 @@ fn test_primitive_magic_methods(source: &str, mapping: &HashMap<&str, &str>) {
     println!("source:\n{source}");
     let mut env = TestEnvironment::basic_test_env();
     let id_to_name = std::mem::take(&mut env.id_to_name);
-    let mut defined_identifiers: HashSet<_> = env.identifier_mapping.keys().copied().collect();
-    defined_identifiers.insert("virtual".into());
+    let mut defined_identifiers: HashMap<_, _> =
+        env.identifier_mapping.keys().copied().map(|id| (id, IdentifierInfo::default())).collect();
+    defined_identifiers.insert("virtual".into(), IdentifierInfo::default());
     let mut inferencer = env.get_inferencer();
     inferencer.defined_identifiers.clone_from(&defined_identifiers);
     let statements = parse_program(source, FileName::default()).unwrap();

nac3core/src/typecheck/typedef/mod.rs

@@ -1,23 +1,28 @@
+use std::{
+    borrow::Cow,
+    cell::RefCell,
+    collections::{HashMap, HashSet},
+    fmt::{self, Display},
+    iter::{repeat, zip},
+    rc::Rc,
+    sync::{Arc, Mutex},
+};
+
 use indexmap::IndexMap;
-use itertools::Itertools;
-use std::cell::RefCell;
-use std::collections::HashMap;
-use std::fmt::{self, Display};
-use std::iter::zip;
-use std::rc::Rc;
-use std::sync::{Arc, Mutex};
-use std::{borrow::Cow, collections::HashSet};
+use itertools::{repeat_n, Itertools};
 
 use nac3parser::ast::{Cmpop, Location, StrRef, Unaryop};
 
-use super::magic_methods::Binop;
-use super::type_error::{TypeError, TypeErrorKind};
-use super::unification_table::{UnificationKey, UnificationTable};
-use crate::symbol_resolver::SymbolValue;
-use crate::toplevel::helper::PrimDef;
-use crate::toplevel::{DefinitionId, TopLevelContext, TopLevelDef};
-use crate::typecheck::magic_methods::OpInfo;
-use crate::typecheck::type_inferencer::PrimitiveStore;
+use super::{
+    magic_methods::{Binop, HasOpInfo, OpInfo},
+    type_error::{TypeError, TypeErrorKind},
+    type_inferencer::PrimitiveStore,
+    unification_table::{UnificationKey, UnificationTable},
+};
+use crate::{
+    symbol_resolver::SymbolValue,
+    toplevel::{helper::PrimDef, DefinitionId, TopLevelContext, TopLevelDef},
+};
 
 #[cfg(test)]
 mod test;
@@ -115,6 +120,7 @@ pub struct FuncArg {
     pub name: StrRef,
     pub ty: Type,
     pub default_value: Option<SymbolValue>,
+    pub is_vararg: bool,
 }
 
 impl FuncArg {
@@ -233,6 +239,12 @@ pub enum TypeEnum {
     TTuple {
         /// The types of elements present in this tuple.
         ty: Vec<Type>,
+
+        /// Whether this tuple is used in a vararg context.
+        ///
+        /// If `true`, `ty` must only contain one type, and the tuple is assumed to contain any
+        /// number of `ty`-typed values.
+        is_vararg_ctx: bool,
     },
 
     /// An object type.
@@ -527,7 +539,7 @@ impl Unifier {
             TypeEnum::TVirtual { ty } => self.get_instantiations(*ty).map(|ty| {
                 ty.iter().map(|&ty| self.add_ty(TypeEnum::TVirtual { ty })).collect_vec()
             }),
-            TypeEnum::TTuple { ty } => {
+            TypeEnum::TTuple { ty, is_vararg_ctx } => {
                 let tuples = ty
                     .iter()
                     .map(|ty| self.get_instantiations(*ty).unwrap_or_else(|| vec![*ty]))
@@ -537,7 +549,12 @@ impl Unifier {
                     None
                 } else {
                     Some(
-                        tuples.into_iter().map(|ty| self.add_ty(TypeEnum::TTuple { ty })).collect(),
+                        tuples
+                            .into_iter()
+                            .map(|ty| {
+                                self.add_ty(TypeEnum::TTuple { ty, is_vararg_ctx: *is_vararg_ctx })
+                            })
+                            .collect(),
                     )
                 }
             }
@@ -581,7 +598,7 @@ impl Unifier {
             TVar { .. } => allowed_typevars.iter().any(|b| self.unification_table.unioned(a, *b)),
             TCall { .. } => false,
             TVirtual { ty } => self.is_concrete(*ty, allowed_typevars),
-            TTuple { ty } => ty.iter().all(|ty| self.is_concrete(*ty, allowed_typevars)),
+            TTuple { ty, .. } => ty.iter().all(|ty| self.is_concrete(*ty, allowed_typevars)),
             TObj { params: vars, .. } => {
                 vars.values().all(|ty| self.is_concrete(*ty, allowed_typevars))
             }
@@ -649,6 +666,7 @@ impl Unifier {
 
         // Get details about the function signature/parameters.
         let num_params = signature.args.len();
+        let is_vararg = signature.args.iter().any(|arg| arg.is_vararg);
 
         // Force the type vars in `b` and `signature' to be up-to-date.
         let b = self.instantiate_fun(b, signature);
@@ -659,8 +677,8 @@ impl Unifier {
         let num_args = posargs.len() + kwargs.len();
 
         // Now we check the arguments against the parameters,
-        // and depending on what `call_info` is, we might change how the behavior `unify_call()`
-        // in hopes to improve user error messages when type checking fails.
+        // and depending on what `call_info` is, we might change how `unify_call()` behaves
+        // to improve user error messages when type checking fails.
         match operator_info {
             Some(OperatorInfo::IsBinaryOp { self_type, operator }) => {
                 // The call is written in the form of (say) `a + b`.
@@ -737,7 +755,7 @@ impl Unifier {
                 };
 
                 // Check for "too many arguments"
-                if num_params < posargs.len() {
+                if !is_vararg && num_params < posargs.len() {
                     let expected_min_count =
                         signature.args.iter().filter(|param| param.is_required()).count();
                     let expected_max_count = num_params;
@@ -770,6 +788,19 @@ impl Unifier {
                     type_check_arg(param.name, param.ty, arg_ty)?;
                 }
 
+                if is_vararg {
+                    debug_assert!(!signature.args.is_empty());
+
+                    let vararg_args = posargs.iter().skip(signature.args.len());
+                    let vararg_param = signature.args.last().unwrap();
+
+                    for (&arg_ty, param) in zip(vararg_args, repeat(vararg_param)) {
+                        // `param_info` for this argument would've already been marked as supplied
+                        // during non-vararg posarg typecheck
+                        type_check_arg(param.name, param.ty, arg_ty)?;
+                    }
+                }
+
                 // Now consume all keyword arguments and typecheck them.
                 for (&param_name, &arg_ty) in kwargs {
                     // We will also use this opportunity to check if this keyword argument is "legal".
@@ -959,7 +990,10 @@ impl Unifier {
                 self.unify_impl(x, b, false)?;
                 self.set_a_to_b(a, x);
             }
-            (TVar { fields: Some(fields), range, is_const_generic: false, .. }, TTuple { ty }) => {
+            (
+                TVar { fields: Some(fields), range, is_const_generic: false, .. },
+                TTuple { ty, .. },
+            ) => {
                 let len = i32::try_from(ty.len()).unwrap();
                 for (k, v) in fields {
                     match *k {
@@ -980,8 +1014,18 @@ impl Unifier {
                             self.unify_impl(v.ty, ty[ind as usize], false)
                                 .map_err(|e| e.at(v.loc))?;
                         }
-                        RecordKey::Str(_) => {
-                            return Err(TypeError::new(TypeErrorKind::NoSuchField(*k, b), v.loc))
+                        RecordKey::Str(s) => {
+                            let tuple_fns = [
+                                Cmpop::Eq.op_info().method_name,
+                                Cmpop::NotEq.op_info().method_name,
+                            ];
+
+                            if !tuple_fns.into_iter().any(|op| s.to_string() == op) {
+                                return Err(TypeError::new(
+                                    TypeErrorKind::NoSuchField(*k, b),
+                                    v.loc,
+                                ));
+                            }
                         }
                     }
                 }
@@ -1056,15 +1100,47 @@ impl Unifier {
                 self.set_a_to_b(a, b);
             }
 
-            (TTuple { ty: ty1 }, TTuple { ty: ty2 }) => {
-                if ty1.len() != ty2.len() {
-                    return Err(TypeError::new(TypeErrorKind::IncompatibleTypes(a, b), None));
+            (
+                TTuple { ty: ty1, is_vararg_ctx: is_vararg1 },
+                TTuple { ty: ty2, is_vararg_ctx: is_vararg2 },
+            ) => {
+                // Rules for Tuples:
+                // - ty1: is_vararg && ty2: is_vararg -> ty1[0] == ty2[0]
+                // - ty1: is_vararg && ty2: !is_vararg -> type error (not enough info to infer the correct number of arguments)
+                // - ty1: !is_vararg && ty2: is_vararg -> ty1[..] == ty2[0]
+                // - ty1: !is_vararg && ty2: !is_vararg -> ty1.len() == ty2.len() && ty1[i] == ty2[i]
+
+                debug_assert!(!is_vararg1 || ty1.len() == 1);
+                debug_assert!(!is_vararg2 || ty2.len() == 1);
+
+                match (*is_vararg1, *is_vararg2) {
+                    (true, true) => {
+                        if self.unify_impl(ty1[0], ty2[0], false).is_err() {
+                            return Self::incompatible_types(a, b);
                         }
+                    }
+                    (true, false) => return Self::incompatible_types(a, b),
+
+                    (false, true) => {
+                        for y in ty2 {
+                            if self.unify_impl(ty1[0], *y, false).is_err() {
+                                return Self::incompatible_types(a, b);
+                            }
+                        }
+                    }
+                    (false, false) => {
+                        if ty1.len() != ty2.len() {
+                            return Self::incompatible_types(a, b);
+                        }
+
                         for (x, y) in ty1.iter().zip(ty2.iter()) {
                             if self.unify_impl(*x, *y, false).is_err() {
-                        return Err(TypeError::new(TypeErrorKind::IncompatibleTypes(a, b), None));
+                                return Self::incompatible_types(a, b);
                             }
                         }
+                    }
+                }
+
                 self.set_a_to_b(a, b);
             }
             (TVar { fields: Some(map), range, .. }, TObj { obj_id, fields, params }) => {
@@ -1307,11 +1383,23 @@ impl Unifier {
             TypeEnum::TLiteral { values, .. } => {
                 format!("const({})", values.iter().map(|v| format!("{v:?}")).join(", "))
             }
-            TypeEnum::TTuple { ty } => {
-                let mut fields =
-                    ty.iter().map(|v| self.internal_stringify(*v, obj_to_name, var_to_name, notes));
+            TypeEnum::TTuple { ty, is_vararg_ctx } => {
+                if *is_vararg_ctx {
+                    debug_assert_eq!(ty.len(), 1);
+                    let field = self.internal_stringify(
+                        *ty.iter().next().unwrap(),
+                        obj_to_name,
+                        var_to_name,
+                        notes,
+                    );
+                    format!("tuple[*{field}]")
+                } else {
+                    let mut fields = ty
+                        .iter()
+                        .map(|v| self.internal_stringify(*v, obj_to_name, var_to_name, notes));
                     format!("tuple[{}]", fields.join(", "))
                 }
+            }
             TypeEnum::TVirtual { ty } => {
                 format!(
                     "virtual[{}]",
@@ -1335,17 +1423,21 @@ impl Unifier {
                     .args
                     .iter()
                     .map(|arg| {
+                        let vararg_prefix = if arg.is_vararg { "*" } else { "" };
+
                         if let Some(dv) = &arg.default_value {
                             format!(
-                                "{}:{}={}",
+                                "{}:{}{}={}",
                                 arg.name,
+                                vararg_prefix,
                                 self.internal_stringify(arg.ty, obj_to_name, var_to_name, notes),
                                 dv
                             )
                         } else {
                             format!(
-                                "{}:{}",
+                                "{}:{}{}",
                                 arg.name,
+                                vararg_prefix,
                                 self.internal_stringify(arg.ty, obj_to_name, var_to_name, notes)
                             )
                         }
@@ -1431,7 +1523,7 @@ impl Unifier {
         match &*ty {
             TypeEnum::TRigidVar { .. } | TypeEnum::TLiteral { .. } => None,
             TypeEnum::TVar { id, .. } => mapping.get(id).copied(),
-            TypeEnum::TTuple { ty } => {
+            TypeEnum::TTuple { ty, is_vararg_ctx } => {
                 let mut new_ty = Cow::from(ty);
                 for (i, t) in ty.iter().enumerate() {
                     if let Some(t1) = self.subst_impl(*t, mapping, cache) {
@@ -1439,7 +1531,10 @@ impl Unifier {
                     }
                 }
                 if matches!(new_ty, Cow::Owned(_)) {
-                    Some(self.add_ty(TypeEnum::TTuple { ty: new_ty.into_owned() }))
+                    Some(self.add_ty(TypeEnum::TTuple {
+                        ty: new_ty.into_owned(),
+                        is_vararg_ctx: *is_vararg_ctx,
+                    }))
                 } else {
                     None
                 }
@@ -1599,16 +1694,37 @@ impl Unifier {
                 }
             }
             (TVar { range, .. }, _) => self.check_var_compatibility(b, range).or(Err(())),
-            (TTuple { ty: ty1 }, TTuple { ty: ty2 }) if ty1.len() == ty2.len() => {
-                let ty: Vec<_> = zip(ty1.iter(), ty2.iter())
-                    .map(|(a, b)| self.get_intersection(*a, *b))
-                    .try_collect()?;
-                if ty.iter().any(Option::is_some) {
-                    Ok(Some(self.add_ty(TTuple {
-                        ty: zip(ty, ty1.iter()).map(|(a, b)| a.unwrap_or(*b)).collect(),
-                    })))
+            (
+                TTuple { ty: ty1, is_vararg_ctx: is_vararg1 },
+                TTuple { ty: ty2, is_vararg_ctx: is_vararg2 },
+            ) => {
+                if *is_vararg1 && *is_vararg2 {
+                    let isect_ty = self.get_intersection(ty1[0], ty2[0])?;
+                    Ok(isect_ty.map(|ty| self.add_ty(TTuple { ty: vec![ty], is_vararg_ctx: true })))
                 } else {
-                    Ok(None)
+                    let zip_iter: Box<dyn Iterator<Item = (&Type, &Type)>> =
+                        match (*is_vararg1, *is_vararg2) {
+                            (true, _) => Box::new(repeat_n(&ty1[0], ty2.len()).zip(ty2.iter())),
+                            (_, false) => Box::new(ty1.iter().zip(repeat_n(&ty2[0], ty1.len()))),
+                            _ => {
+                                if ty1.len() != ty2.len() {
+                                    return Err(());
+                                }
+
+                                Box::new(ty1.iter().zip(ty2.iter()))
+                            }
+                        };
+
+                    let ty: Vec<_> =
+                        zip_iter.map(|(a, b)| self.get_intersection(*a, *b)).try_collect()?;
+                    Ok(if ty.iter().any(Option::is_some) {
+                        Some(self.add_ty(TTuple {
+                            ty: zip(ty, ty1.iter()).map(|(a, b)| a.unwrap_or(*b)).collect(),
+                            is_vararg_ctx: false,
+                        }))
+                    } else {
+                        None
+                    })
                 }
             }
             // TODO(Derppening): #444

nac3core/src/typecheck/typedef/test.rs

@@ -1,10 +1,12 @@
-use super::super::magic_methods::with_fields;
-use super::*;
+use std::collections::HashMap;
+
 use indoc::indoc;
 use itertools::Itertools;
-use std::collections::HashMap;
 use test_case::test_case;
 
+use super::*;
+use crate::typecheck::magic_methods::with_fields;
+
 impl Unifier {
     /// Check whether two types are equal.
     fn eq(&mut self, a: Type, b: Type) -> bool {
@@ -28,7 +30,10 @@ impl Unifier {
                 TypeEnum::TVar { fields: Some(map1), .. },
                 TypeEnum::TVar { fields: Some(map2), .. },
             ) => self.map_eq2(map1, map2),
-            (TypeEnum::TTuple { ty: ty1 }, TypeEnum::TTuple { ty: ty2 }) => {
+            (
+                TypeEnum::TTuple { ty: ty1, is_vararg_ctx: false },
+                TypeEnum::TTuple { ty: ty2, is_vararg_ctx: false },
+            ) => {
                 ty1.len() == ty2.len()
                     && ty1.iter().zip(ty2.iter()).all(|(t1, t2)| self.eq(*t1, *t2))
             }
@@ -178,7 +183,7 @@ impl TestEnvironment {
                     ty.push(result.0);
                     s = result.1;
                 }
-                (self.unifier.add_ty(TypeEnum::TTuple { ty }), &s[1..])
+                (self.unifier.add_ty(TypeEnum::TTuple { ty, is_vararg_ctx: false }), &s[1..])
             }
             "Record" => {
                 let mut s = &typ[end..];
@@ -608,7 +613,7 @@ fn test_instantiation() {
     let v1 = env.unifier.get_fresh_var_with_range(&[list_v, int], None, None).ty;
     let v2 = env.unifier.get_fresh_var_with_range(&[list_int, float], None, None).ty;
     let t = env.unifier.get_dummy_var().ty;
-    let tuple = env.unifier.add_ty(TypeEnum::TTuple { ty: vec![v, v1, v2] });
+    let tuple = env.unifier.add_ty(TypeEnum::TTuple { ty: vec![v, v1, v2], is_vararg_ctx: false });
     let v3 = env.unifier.get_fresh_var_with_range(&[tuple, t], None, None).ty;
     // t = TypeVar('t')
     // v = TypeVar('v', int, bool)

nac3ld/src/dwarf.rs

@@ -238,7 +238,7 @@ impl<'a> EH_Frame<'a> {
 /// From the [specification](https://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/ehframechpt.html):
 ///
 /// > Each CFI record contains a Common Information Entry (CIE) record followed by 1 or more Frame
-/// Description Entry (FDE) records.
+/// > Description Entry (FDE) records.
 pub struct CFI_Record<'a> {
     // It refers to the augmentation data that corresponds to 'R' in the augmentation string
     fde_pointer_encoding: u8,

nac3ld/src/lib.rs

@@ -1,10 +1,4 @@
-#![deny(
-    future_incompatible,
-    let_underscore,
-    nonstandard_style,
-    rust_2024_compatibility,
-    clippy::all
-)]
+#![deny(future_incompatible, let_underscore, nonstandard_style, clippy::all)]
 #![warn(clippy::pedantic)]
 #![allow(
     clippy::cast_possible_truncation,
@@ -21,13 +15,12 @@
     clippy::wildcard_imports
 )]
 
+use std::{collections::HashMap, mem, ptr, slice, str};
+
+use byteorder::{ByteOrder, LittleEndian};
+
 use dwarf::*;
 use elf::*;
-use std::collections::HashMap;
-use std::{mem, ptr, slice, str};
-
-extern crate byteorder;
-use byteorder::{ByteOrder, LittleEndian};
 
 mod dwarf;
 mod elf;

nac3parser/Cargo.toml

@@ -8,15 +8,15 @@ license = "MIT"
 edition = "2021"
 
 [build-dependencies]
-lalrpop = "0.20"
+lalrpop = "0.22"
 
 [dependencies]
 nac3ast = { path = "../nac3ast" }
-lalrpop-util = "0.20"
+lalrpop-util = "0.22"
 log = "0.4"
 unic-emoji-char = "0.9"
 unic-ucd-ident  = "0.9"
-unicode_names2 = "1.2"
+unicode_names2 = "1.3"
 phf = { version = "0.11", features = ["macros"] }
 ahash = "0.8"
 

nac3parser/src/config_comment_helper.rs

@@ -1,8 +1,10 @@
-use crate::ast::Ident;
-use crate::ast::Location;
-use crate::error::*;
-use crate::token::Tok;
+use crate::{
+    ast::{Ident, Location},
+    error::*,
+    token::Tok,
+};
 use lalrpop_util::ParseError;
+
 use nac3ast::*;
 
 pub fn make_config_comment(

nac3parser/src/error.rs

@@ -1,13 +1,12 @@
 //! Define internal parse error types
 //! The goal is to provide a matching and a safe error API, maksing errors from LALR
-use lalrpop_util::ParseError as LalrpopError;
-
-use crate::ast::Location;
-use crate::token::Tok;
-
 use std::error::Error;
 use std::fmt;
 
+use lalrpop_util::ParseError as LalrpopError;
+
+use crate::{ast::Location, token::Tok};
+
 /// Represents an error during lexical scanning.
 #[derive(Debug, PartialEq)]
 pub struct LexicalError {