WIP: core: remove outdated comment

Details:
- improve irrt model
- len() on ndarrays

.clang-format

@@ -1,32 +0,0 @@
-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

@@ -1,4 +1,3 @@
 __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: [pre-commit]
+default_stages: [commit]
 
 repos:
   - repo: local
     hooks:
       - id: nac3-cargo-fmt
         name: nac3 cargo format
-        entry: nix
+        entry: cargo
         language: system
         types: [file, rust]
         pass_filenames: false
         description: Runs cargo fmt on the codebase.
-        args: [develop, -c, cargo, fmt, --all]
+        args: [fmt]
       - id: nac3-cargo-clippy
         name: nac3 cargo clippy
-        entry: nix
+        entry: cargo
         language: system
         types: [file, rust]
         pass_filenames: false
         description: Runs cargo clippy on the codebase.
-        args: [develop, -c, cargo, clippy, --tests]
+        args: [clippy, --tests]

Cargo.toml

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

flake.lock

@@ -2,11 +2,11 @@
   "nodes": {
     "nixpkgs": {
       "locked": {
-        "lastModified": 1736798957,
-        "narHash": "sha256-qwpCtZhSsSNQtK4xYGzMiyEDhkNzOCz/Vfu4oL2ETsQ=",
+        "lastModified": 1720418205,
+        "narHash": "sha256-cPJoFPXU44GlhWg4pUk9oUPqurPlCFZ11ZQPk21GTPU=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "9abb87b552b7f55ac8916b6fc9e5cb486656a2f3",
+        "rev": "655a58a72a6601292512670343087c2d75d859c1",
         "type": "github"
       },
       "original": {

flake.nix

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

nac3artiq/Cargo.toml

@@ -12,10 +12,15 @@ crate-type = ["cdylib"]
 itertools = "0.13"
 pyo3 = { version = "0.21", features = ["extension-module", "gil-refs"] }
 parking_lot = "0.12"
-tempfile = "3.13"
+tempfile = "3.10"
+nac3parser = { path = "../nac3parser" }
 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/embedding_map.py

@@ -0,0 +1,66 @@
+class EmbeddingMap:
+    def __init__(self):
+        self.object_inverse_map = {}
+        self.object_map = {}
+        self.string_map = {}
+        self.string_reverse_map = {}
+        self.function_map = {}
+        self.attributes_writeback = []
+
+        # preallocate exception names
+        self.preallocate_runtime_exception_names(["RuntimeError",
+                                          "RTIOUnderflow",
+                                          "RTIOOverflow",
+                                          "RTIODestinationUnreachable",
+                                          "DMAError",
+                                          "I2CError",
+                                          "CacheError",
+                                          "SPIError",
+                                          "0:ZeroDivisionError",
+                                          "0:IndexError",
+                                          "0:ValueError",
+                                          "0:RuntimeError",
+                                          "0:AssertionError",
+                                          "0:KeyError",
+                                          "0:NotImplementedError",
+                                          "0:OverflowError",
+                                          "0:IOError",
+                                          "0:UnwrapNoneError"])
+
+    def preallocate_runtime_exception_names(self, names):
+        for i, name in enumerate(names):
+            if ":" not in name:
+                name = "0:artiq.coredevice.exceptions." + name
+            exn_id = self.store_str(name)
+            assert exn_id == i
+
+    def store_function(self, key, fun):
+        self.function_map[key] = fun
+        return key
+
+    def store_object(self, obj):
+        obj_id = id(obj)
+        if obj_id in self.object_inverse_map:
+            return self.object_inverse_map[obj_id]
+        key = len(self.object_map) + 1
+        self.object_map[key] = obj
+        self.object_inverse_map[obj_id] = key
+        return key
+
+    def store_str(self, s):
+        if s in self.string_reverse_map:
+            return self.string_reverse_map[s]
+        key = len(self.string_map)
+        self.string_map[key] = s
+        self.string_reverse_map[s] = key
+        return key
+
+    def retrieve_function(self, key):
+        return self.function_map[key]
+
+    def retrieve_object(self, key):
+        return self.object_map[key]
+
+    def retrieve_str(self, key):
+        return self.string_map[key]
+

nac3artiq/demo/min_artiq.py

@@ -6,6 +6,7 @@ from typing import Generic, TypeVar
 from math import floor, ceil
 
 import nac3artiq
+from embedding_map import EmbeddingMap
 
 
 __all__ = [
@@ -111,15 +112,10 @@ def extern(function):
     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 rpc(function):
+    """Decorates a function declaration defined by the core device runtime."""
+    register_function(function)
+    return function
 
 def kernel(function_or_method):
     """Decorates a function or method to be executed on the core device."""
@@ -192,46 +188,6 @@ def print_int64(x: int64):
     raise NotImplementedError("syscall not simulated")
 
 
-class EmbeddingMap:
-    def __init__(self):
-        self.object_inverse_map = {}
-        self.object_map = {}
-        self.string_map = {}
-        self.string_reverse_map = {}
-        self.function_map = {}
-        self.attributes_writeback = []
-
-    def store_function(self, key, fun):
-        self.function_map[key] = fun
-        return key
-
-    def store_object(self, obj):
-        obj_id = id(obj)
-        if obj_id in self.object_inverse_map:
-            return self.object_inverse_map[obj_id]
-        key = len(self.object_map) + 1
-        self.object_map[key] = obj
-        self.object_inverse_map[obj_id] = key
-        return key
-
-    def store_str(self, s):
-        if s in self.string_reverse_map:
-            return self.string_reverse_map[s]
-        key = len(self.string_map)
-        self.string_map[key] = s
-        self.string_reverse_map[s] = key
-        return key
-
-    def retrieve_function(self, key):
-        return self.function_map[key]
-
-    def retrieve_object(self, key):
-        return self.object_map[key]
-
-    def retrieve_str(self, key):
-        return self.string_map[key]
-
-
 @nac3
 class Core:
     ref_period: KernelInvariant[float]
@@ -245,7 +201,7 @@ class Core:
         embedding = EmbeddingMap()
 
         if allow_registration:
-            compiler.analyze(registered_functions, registered_classes, set())
+            compiler.analyze(registered_functions, registered_classes)
             allow_registration = False
 
         if hasattr(method, "__self__"):

nac3artiq/demo/module.py

@@ -1,26 +0,0 @@
-from min_artiq import *
-from numpy import int32
-
-# Global Variable Definition
-X: Kernel[int32] = 1
-
-# TopLevelFunction Defintion
-@kernel
-def display_X():
-    print_int32(X)
-
-# TopLevel Class Definition
-@nac3
-class A:    
-    @kernel
-    def __init__(self):
-        self.set_x(1)
-    
-    @kernel
-    def set_x(self, new_val: int32):
-        global X
-        X = new_val
-    
-    @kernel
-    def get_X(self) -> int32:
-        return X

nac3artiq/demo/module_support.py

@@ -1,26 +0,0 @@
-from min_artiq import *
-import module as module_definition
-
-@nac3
-class TestModuleSupport:
-    core: KernelInvariant[Core]
-
-    def __init__(self):
-        self.core = Core()
-
-    @kernel
-    def run(self):
-        # Accessing classes
-        obj = module_definition.A()
-        obj.get_X()
-        obj.set_x(2)
-        
-        # Calling functions
-        module_definition.display_X()
-
-        # Updating global variables
-        module_definition.X = 9
-        module_definition.display_X()
-
-if __name__ == "__main__":
-    TestModuleSupport().run()

nac3artiq/demo/numpy_primitives_decay.py

@@ -1,29 +0,0 @@
-from min_artiq import *
-import numpy
-from numpy import int32
-
-
-@nac3
-class NumpyBoolDecay:
-    core: KernelInvariant[Core]
-    np_true: KernelInvariant[bool]
-    np_false: KernelInvariant[bool]
-    np_int: KernelInvariant[int32]
-    np_float: KernelInvariant[float]
-    np_str: KernelInvariant[str]
-
-    def __init__(self):
-        self.core = Core()
-        self.np_true = numpy.True_
-        self.np_false = numpy.False_
-        self.np_int = numpy.int32(0)
-        self.np_float = numpy.float64(0.0)
-        self.np_str = numpy.str_("")
-
-    @kernel
-    def run(self):
-        pass
-
-
-if __name__ == "__main__":
-    NumpyBoolDecay().run()

nac3artiq/demo/str_abi.py

@@ -1,26 +0,0 @@
-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/demo/string_attribute_issue337.py

@@ -3,22 +3,22 @@ from numpy import int32
 
 
 @nac3
-class EmptyList:
+class Demo:
     core: KernelInvariant[Core]
+    attr1: KernelInvariant[str]
+    attr2: KernelInvariant[int32]
+
 
     def __init__(self):
         self.core = Core()
-
-    @rpc
-    def get_empty(self) -> list[int32]:
-        return []
+        self.attr2 = 32
+        self.attr1 = "SAMPLE"
 
     @kernel
     def run(self):
-        a: list[int32] = self.get_empty()
-        if a != []:
-            raise ValueError
+        print_int32(self.attr2)
+        self.attr1
 
 
 if __name__ == "__main__":
-    EmptyList().run()
+    Demo().run()

nac3artiq/demo/support_class_attr_issue102.py

@@ -0,0 +1,40 @@
+from min_artiq import *
+from numpy import int32
+
+
+@nac3
+class Demo:
+    attr1: KernelInvariant[int32] = 2
+    attr2: int32 = 4
+    attr3: Kernel[int32]
+
+    @kernel
+    def __init__(self):
+        self.attr3 = 8
+
+
+@nac3
+class NAC3Devices:
+    core: KernelInvariant[Core]
+    attr4: KernelInvariant[int32] = 16
+    
+    def __init__(self):
+        self.core = Core()
+
+    @kernel
+    def run(self):
+        Demo.attr1  # Supported
+        # Demo.attr2  # Field not accessible on Kernel
+        # Demo.attr3  # Only attributes can be accessed in this way
+        # Demo.attr1 = 2  # Attributes are immutable
+
+        self.attr4  # Attributes can be accessed within class
+
+        obj = Demo()
+        obj.attr1  # Attributes can be accessed by class objects
+
+        NAC3Devices.attr4  # Attributes accessible for classes without __init__
+
+
+if __name__ == "__main__":
+    NAC3Devices().run()

nac3artiq/src/lib.rs

@@ -1,4 +1,10 @@
-#![deny(future_incompatible, let_underscore, nonstandard_style, clippy::all)]
+#![deny(
+    future_incompatible,
+    let_underscore,
+    nonstandard_style,
+    rust_2024_compatibility,
+    clippy::all
+)]
 #![warn(clippy::pedantic)]
 #![allow(
     unsafe_op_in_unsafe_fn,
@@ -10,65 +16,63 @@
     clippy::wildcard_imports
 )]
 
-use std::{
-    collections::{HashMap, HashSet},
-    fs,
-    io::Write,
-    process::Command,
-    rc::Rc,
-    sync::Arc,
-};
+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 itertools::Itertools;
-use parking_lot::{Mutex, RwLock};
-use pyo3::{
-    create_exception, exceptions,
-    prelude::*,
-    types::{PyBytes, PyDict, PyNone, PySet},
+use inkwell::{
+    memory_buffer::MemoryBuffer,
+    module::{Linkage, Module},
+    passes::PassBuilderOptions,
+    support::is_multithreaded,
+    targets::*,
+    OptimizationLevel,
 };
-use tempfile::{self, TempDir};
+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},
+    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::{
-        concrete_type::ConcreteTypeStore, gen_func_impl, irrt::load_irrt, CodeGenLLVMOptions,
-        CodeGenTargetMachineOptions, CodeGenTask, CodeGenerator, WithCall, WorkerRegistry,
-    },
-    inkwell::{
-        context::Context,
-        memory_buffer::MemoryBuffer,
-        module::{FlagBehavior, Linkage, Module},
-        passes::PassBuilderOptions,
-        support::is_multithreaded,
-        targets::*,
-        OptimizationLevel,
-    },
-    nac3parser::{
-        ast::{self, Constant, ExprKind, Located, Stmt, StmtKind, StrRef},
-        parser::parse_program,
-    },
+    codegen::irrt::load_irrt,
+    codegen::{concrete_type::ConcreteTypeStore, CodeGenTask, WithCall, WorkerRegistry},
     symbol_resolver::SymbolResolver,
     toplevel::{
-        builtins::get_exn_constructor,
-        composer::{BuiltinFuncCreator, BuiltinFuncSpec, ComposerConfig, TopLevelComposer},
+        composer::{ComposerConfig, TopLevelComposer},
         DefinitionId, GenCall, TopLevelDef,
     },
-    typecheck::{
-        type_inferencer::PrimitiveStore,
-        typedef::{into_var_map, FunSignature, FuncArg, Type, TypeEnum, Unifier, VarMap},
-    },
+    typecheck::typedef::{FunSignature, FuncArg},
+    typecheck::{type_inferencer::PrimitiveStore, typedef::Type},
 };
+
 use nac3ld::Linker;
 
-use codegen::{
-    attributes_writeback, gen_core_log, gen_rtio_log, rpc_codegen_callback, ArtiqCodeGenerator,
+use tempfile::{self, TempDir};
+
+use crate::codegen::attributes_writeback;
+use crate::{
+    codegen::{rpc_codegen_callback, ArtiqCodeGenerator},
+    symbol_resolver::{DeferredEvaluationStore, InnerResolver, PythonHelper, Resolver},
 };
-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,
@@ -78,62 +82,14 @@ enum Isa {
 }
 
 impl Isa {
-    /// Returns the [`TargetTriple`] used for compiling to this ISA.
-    pub fn get_llvm_target_triple(self) -> TargetTriple {
-        match self {
-            Isa::Host => TargetMachine::get_default_triple(),
-            Isa::RiscV32G | Isa::RiscV32IMA => TargetTriple::create("riscv32-unknown-linux"),
-            Isa::CortexA9 => TargetTriple::create("armv7-unknown-linux-gnueabihf"),
+    /// Returns the number of bits in `size_t` for the [`Isa`].
+    fn get_size_type(self) -> u32 {
+        if self == Isa::Host {
+            64u32
+        } else {
+            32u32
         }
     }
-
-    /// Returns the [`String`] representing the target CPU used for compiling to this ISA.
-    pub fn get_llvm_target_cpu(self) -> String {
-        match self {
-            Isa::Host => TargetMachine::get_host_cpu_name().to_string(),
-            Isa::RiscV32G | Isa::RiscV32IMA => "generic-rv32".to_string(),
-            Isa::CortexA9 => "cortex-a9".to_string(),
-        }
-    }
-
-    /// Returns the [`String`] representing the target features used for compiling to this ISA.
-    pub fn get_llvm_target_features(self) -> String {
-        match self {
-            Isa::Host => TargetMachine::get_host_cpu_features().to_string(),
-            Isa::RiscV32G => "+a,+m,+f,+d".to_string(),
-            Isa::RiscV32IMA => "+a,+m".to_string(),
-            Isa::CortexA9 => "+dsp,+fp16,+neon,+vfp3,+long-calls".to_string(),
-        }
-    }
-
-    /// Returns an instance of [`CodeGenTargetMachineOptions`] representing the target machine
-    /// options used for compiling to this ISA.
-    pub fn get_llvm_target_options(self) -> CodeGenTargetMachineOptions {
-        CodeGenTargetMachineOptions {
-            triple: self.get_llvm_target_triple().as_str().to_string_lossy().into_owned(),
-            cpu: self.get_llvm_target_cpu(),
-            features: self.get_llvm_target_features(),
-            reloc_mode: RelocMode::PIC,
-            ..CodeGenTargetMachineOptions::from_host()
-        }
-    }
-
-    /// Returns an instance of [`TargetMachine`] used in compiling and linking of a program of this
-    /// ISA.
-    pub fn create_llvm_target_machine(self, opt_level: OptimizationLevel) -> TargetMachine {
-        self.get_llvm_target_options()
-            .create_target_machine(opt_level)
-            .expect("couldn't create target machine")
-    }
-
-    /// Returns the number of bits in `size_t` for this ISA.
-    fn get_size_type(self, ctx: &Context) -> u32 {
-        ctx.ptr_sized_int_type(
-            &self.create_llvm_target_machine(OptimizationLevel::Default).get_target_data(),
-            None,
-        )
-        .get_bit_width()
-    }
 }
 
 #[derive(Clone)]
@@ -159,7 +115,6 @@ pub struct PrimitivePythonId {
     generic_alias: (u64, u64),
     virtual_id: u64,
     option: u64,
-    module: u64,
 }
 
 type TopLevelComponent = (Stmt, String, PyObject);
@@ -171,7 +126,7 @@ struct Nac3 {
     isa: Isa,
     time_fns: &'static (dyn TimeFns + Sync),
     primitive: PrimitiveStore,
-    builtins: Vec<BuiltinFuncSpec>,
+    builtins: Vec<(StrRef, FunSignature, Arc<GenCall>)>,
     pyid_to_def: Arc<RwLock<HashMap<u64, DefinitionId>>>,
     primitive_ids: PrimitivePythonId,
     working_directory: TempDir,
@@ -191,32 +146,14 @@ impl Nac3 {
         module: &PyObject,
         registered_class_ids: &HashSet<u64>,
     ) -> PyResult<()> {
-        let (module_name, source_file, source) =
-            Python::with_gil(|py| -> PyResult<(String, String, String)> {
-                let module: &PyAny = module.extract(py)?;
-                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 (module_name, source_file) = Python::with_gil(|py| -> PyResult<(String, String)> {
+            let module: &PyAny = module.extract(py)?;
+            Ok((module.getattr("__name__")?.extract()?, module.getattr("__file__")?.extract()?))
+        })?;
 
+        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}")))?;
 
@@ -256,8 +193,10 @@ impl Nac3 {
                     body.retain(|stmt| {
                         if let StmtKind::FunctionDef { ref decorator_list, .. } = stmt.node {
                             decorator_list.iter().any(|decorator| {
-                                if let Some(id) = decorator_id_string(decorator) {
-                                    id == "kernel" || id == "portable" || id == "rpc"
+                                if let ExprKind::Name { id, .. } = decorator.node {
+                                    id.to_string() == "kernel"
+                                        || id.to_string() == "portable"
+                                        || id.to_string() == "rpc"
                                 } else {
                                     false
                                 }
@@ -270,17 +209,14 @@ impl Nac3 {
                 }
                 StmtKind::FunctionDef { ref decorator_list, .. } => {
                     decorator_list.iter().any(|decorator| {
-                        if let Some(id) = decorator_id_string(decorator) {
-                            id == "extern" || id == "kernel" || id == "portable" || id == "rpc"
+                        if let ExprKind::Name { id, .. } = decorator.node {
+                            let id = id.to_string();
+                            id == "extern" || id == "portable" || id == "kernel" || id == "rpc"
                         } else {
                             false
                         }
                     })
                 }
-                // Allow global variable declaration with `Kernel` type annotation
-                StmtKind::AnnAssign { ref annotation, .. } => {
-                    matches!(&annotation.node, ExprKind::Subscript { value, .. } if matches!(&value.node, ExprKind::Name {id, ..} if id == &"Kernel".into()))
-                }
                 _ => false,
             };
 
@@ -328,7 +264,7 @@ 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() => {
@@ -364,64 +300,6 @@ 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.as_ref(), 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.as_ref(), fun, &args, generator)?;
-
-                        Ok(None)
-                    }))),
-                )
-            }),
-        ]
-    }
-
     fn compile_method<T>(
         &self,
         obj: &PyAny,
@@ -431,10 +309,9 @@ impl Nac3 {
         py: Python,
         link_fn: &dyn Fn(&Module) -> PyResult<T>,
     ) -> PyResult<T> {
-        let size_t = self.isa.get_size_type(&Context::create());
+        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,
         );
@@ -474,14 +351,12 @@ impl Nac3 {
         ];
         add_exceptions(&mut composer, &mut builtins_def, &mut builtins_ty, &exception_names);
 
-        // Stores a mapping from module id to attributes
         let mut module_to_resolver_cache: HashMap<u64, _> = HashMap::new();
 
         let mut rpc_ids = vec![];
         for (stmt, path, module) in &self.top_levels {
             let py_module: &PyAny = module.extract(py)?;
             let module_id: u64 = id_fn.call1((py_module,))?.extract()?;
-            let module_name: String = py_module.getattr("__name__")?.extract()?;
             let helper = helper.clone();
             let class_obj;
             if let StmtKind::ClassDef { name, .. } = &stmt.node {
@@ -496,7 +371,7 @@ impl Nac3 {
             } else {
                 class_obj = None;
             }
-            let (name_to_pyid, resolver, _, _) =
+            let (name_to_pyid, resolver) =
                 module_to_resolver_cache.get(&module_id).cloned().unwrap_or_else(|| {
                     let mut name_to_pyid: HashMap<StrRef, u64> = HashMap::new();
                     let members: &PyDict =
@@ -513,6 +388,7 @@ 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(),
@@ -525,17 +401,9 @@ impl Nac3 {
                     })))
                         as Arc<dyn SymbolResolver + Send + Sync>;
                     let name_to_pyid = Rc::new(name_to_pyid);
-                    let module_location = ast::Location::new(1, 1, stmt.location.file);
-                    module_to_resolver_cache.insert(
-                        module_id,
-                        (
-                            name_to_pyid.clone(),
-                            resolver.clone(),
-                            module_name.clone(),
-                            Some(module_location),
-                        ),
-                    );
-                    (name_to_pyid, resolver, module_name, Some(module_location))
+                    module_to_resolver_cache
+                        .insert(module_id, (name_to_pyid.clone(), resolver.clone()));
+                    (name_to_pyid, resolver)
                 });
 
             let (name, def_id, ty) = composer
@@ -551,25 +419,9 @@ impl Nac3 {
 
             match &stmt.node {
                 StmtKind::FunctionDef { decorator_list, .. } => {
-                    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));
+                    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));
                     }
                 }
                 StmtKind::ClassDef { name, body, .. } => {
@@ -577,26 +429,19 @@ 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| {
-                                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 decorator_list.iter().any(|decorator| matches!(decorator.node, ExprKind::Name { id, .. } if id == "rpc".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, is_async));
+                                rpc_ids.push((Some((class_obj.clone(), *name)), def_id));
                             }
                         }
                     }
                 }
-                _ => (),
+                _ => ()
             }
 
             let id = *name_to_pyid.get(&name).unwrap();
@@ -609,24 +454,6 @@ impl Nac3 {
             }
         }
 
-        // Adding top level module definitions
-        for (module_id, (module_name_to_pyid, module_resolver, module_name, module_location)) in
-            module_to_resolver_cache
-        {
-            let def_id = composer
-                .register_top_level_module(
-                    &module_name,
-                    &module_name_to_pyid,
-                    module_resolver,
-                    module_location,
-                )
-                .map_err(|e| {
-                    CompileError::new_err(format!("compilation failed\n----------\n{e}"))
-                })?;
-
-            self.pyid_to_def.write().insert(module_id, def_id);
-        }
-
         let id_fun = PyModule::import(py, "builtins")?.getattr("id")?;
         let mut name_to_pyid: HashMap<StrRef, u64> = HashMap::new();
         let module = PyModule::new(py, "tmp")?;
@@ -653,12 +480,13 @@ 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(),
             name_to_pyid,
             module: module.to_object(py),
-            helper: helper.clone(),
+            helper,
             string_store: self.string_store.clone(),
             exception_ids: self.exception_ids.clone(),
             deferred_eval_store: self.deferred_eval_store.clone(),
@@ -669,10 +497,6 @@ 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();
@@ -710,12 +534,13 @@ 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, is_async) in &rpc_ids {
+            for (class_data, id) in &rpc_ids {
                 let mut def = defs[id.0].write();
                 match &mut *def {
                     TopLevelDef::Function { codegen_callback, .. } => {
-                        *codegen_callback = Some(rpc_codegen_callback(*is_async));
+                        *codegen_callback = Some(rpc_codegen.clone());
                     }
                     TopLevelDef::Class { methods, .. } => {
                         let (class_def, method_name) = class_data.as_ref().unwrap();
@@ -726,7 +551,7 @@ impl Nac3 {
                             if let TopLevelDef::Function { codegen_callback, .. } =
                                 &mut *defs[id.0].write()
                             {
-                                *codegen_callback = Some(rpc_codegen_callback(*is_async));
+                                *codegen_callback = Some(rpc_codegen.clone());
                                 store_fun
                                     .call1(
                                         py,
@@ -741,14 +566,6 @@ impl Nac3 {
                             }
                         }
                     }
-                    TopLevelDef::Variable { .. } => {
-                        return Err(CompileError::new_err(String::from(
-                            "Unsupported @rpc annotation on global variable",
-                        )))
-                    }
-                    TopLevelDef::Module { .. } => {
-                        unreachable!("Type module cannot be decorated with @rpc")
-                    }
                 }
             }
         }
@@ -769,12 +586,33 @@ 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: instance.calls,
+            calls: Arc::new(HashMap::default()),
             id: 0,
         };
 
@@ -787,47 +625,25 @@ impl Nac3 {
             let buffer = buffer.as_slice().into();
             membuffer.lock().push(buffer);
         })));
+        let size_t = if self.isa == Isa::Host { 64 } else { 32 };
         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
             .iter()
-            .map(|s| {
-                Box::new(ArtiqCodeGenerator::with_target_machine(
-                    s.to_string(),
-                    &context,
-                    &self.get_llvm_target_machine(),
-                    self.time_fns,
-                ))
-            })
+            .map(|s| Box::new(ArtiqCodeGenerator::new(s.to_string(), size_t, self.time_fns)))
             .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 context = Context::create();
-            let mut generator = ArtiqCodeGenerator::with_target_machine(
-                "main".to_string(),
-                &context,
-                &self.get_llvm_target_machine(),
-                self.time_fns,
-            );
-            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 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 builder = context.create_builder();
             let (_, module, _) = gen_func_impl(
                 &context,
@@ -835,27 +651,9 @@ impl Nac3 {
                 &registry,
                 builder,
                 module,
-                task,
+                attributes_writeback_task,
                 |generator, ctx| {
-                    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,
-                    )
+                    attributes_writeback(ctx, generator, inner_resolver.as_ref(), &host_attributes)
                 },
             )
             .unwrap();
@@ -864,24 +662,37 @@ impl Nac3 {
             membuffer.lock().push(buffer);
         });
 
-        embedding_map.setattr("expects_return", has_return).unwrap();
-
-        // Link all modules into `main`.
+        let context = inkwell::context::Context::create();
         let buffers = membuffers.lock();
         let main = context
-            .create_module_from_ir(MemoryBuffer::create_from_memory_range(
-                buffers.last().unwrap(),
-                "main",
-            ))
+            .create_module_from_ir(MemoryBuffer::create_from_memory_range(&buffers[0], "main"))
             .unwrap();
-        for buffer in buffers.iter().rev().skip(1) {
+        for buffer in buffers.iter().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()))?;
         }
-        main.link_in_module(irrt).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()))?;
 
         let mut function_iter = main.get_first_function();
         while let Some(func) = function_iter {
@@ -915,65 +726,58 @@ impl Nac3 {
             panic!("Failed to run optimization for module `main`: {}", err.to_string());
         }
 
-        Python::with_gil(|py| {
-            let string_store = self.string_store.read();
-            let mut string_store_vec = string_store.iter().collect::<Vec<_>>();
-            string_store_vec.sort_by(|(_s1, key1), (_s2, key2)| key1.cmp(key2));
-            for (s, key) in string_store_vec {
-                let embed_key: i32 = helper.store_str.call1(py, (s,)).unwrap().extract(py).unwrap();
-                assert_eq!(
-                    embed_key, *key,
-                    "string {s} is out of sync between embedding map (key={embed_key}) and \
-                    the internal string store (key={key})"
-                );
-            }
-        });
-
         link_fn(&main)
     }
 
+    /// Returns the [`TargetTriple`] used for compiling to [isa].
+    fn get_llvm_target_triple(isa: Isa) -> TargetTriple {
+        match isa {
+            Isa::Host => TargetMachine::get_default_triple(),
+            Isa::RiscV32G | Isa::RiscV32IMA => TargetTriple::create("riscv32-unknown-linux"),
+            Isa::CortexA9 => TargetTriple::create("armv7-unknown-linux-gnueabihf"),
+        }
+    }
+
+    /// Returns the [`String`] representing the target CPU used for compiling to [isa].
+    fn get_llvm_target_cpu(isa: Isa) -> String {
+        match isa {
+            Isa::Host => TargetMachine::get_host_cpu_name().to_string(),
+            Isa::RiscV32G | Isa::RiscV32IMA => "generic-rv32".to_string(),
+            Isa::CortexA9 => "cortex-a9".to_string(),
+        }
+    }
+
+    /// Returns the [`String`] representing the target features used for compiling to [isa].
+    fn get_llvm_target_features(isa: Isa) -> String {
+        match isa {
+            Isa::Host => TargetMachine::get_host_cpu_features().to_string(),
+            Isa::RiscV32G => "+a,+m,+f,+d".to_string(),
+            Isa::RiscV32IMA => "+a,+m".to_string(),
+            Isa::CortexA9 => "+dsp,+fp16,+neon,+vfp3,+long-calls".to_string(),
+        }
+    }
+
+    /// Returns an instance of [`CodeGenTargetMachineOptions`] representing the target machine
+    /// options used for compiling to [isa].
+    fn get_llvm_target_options(isa: Isa) -> CodeGenTargetMachineOptions {
+        CodeGenTargetMachineOptions {
+            triple: Nac3::get_llvm_target_triple(isa).as_str().to_string_lossy().into_owned(),
+            cpu: Nac3::get_llvm_target_cpu(isa),
+            features: Nac3::get_llvm_target_features(isa),
+            reloc_mode: RelocMode::PIC,
+            ..CodeGenTargetMachineOptions::from_host()
+        }
+    }
+
     /// Returns an instance of [`TargetMachine`] used in compiling and linking of a program to the
-    /// target [ISA][isa].
+    /// target [isa].
     fn get_llvm_target_machine(&self) -> TargetMachine {
-        self.isa.create_llvm_target_machine(self.llvm_options.opt_level)
+        Nac3::get_llvm_target_options(self.isa)
+            .create_target_machine(self.llvm_options.opt_level)
+            .expect("couldn't create target machine")
     }
 }
 
-/// 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(),
@@ -1043,8 +847,7 @@ impl Nac3 {
             Isa::RiscV32IMA => &timeline::NOW_PINNING_TIME_FNS,
             Isa::CortexA9 | Isa::Host => &timeline::EXTERN_TIME_FNS,
         };
-        let (primitive, _) =
-            TopLevelComposer::make_primitives(isa.get_size_type(&Context::create()));
+        let primitive: PrimitiveStore = TopLevelComposer::make_primitives(isa.get_size_type()).0;
         let builtins = vec![
             (
                 "now_mu".into(),
@@ -1060,7 +863,6 @@ impl Nac3 {
                         name: "t".into(),
                         ty: primitive.int64,
                         default_value: None,
-                        is_vararg: false,
                     }],
                     ret: primitive.none,
                     vars: VarMap::new(),
@@ -1080,7 +882,6 @@ impl Nac3 {
                         name: "dt".into(),
                         ty: primitive.int64,
                         default_value: None,
-                        is_vararg: false,
                     }],
                     ret: primitive.none,
                     vars: VarMap::new(),
@@ -1132,54 +933,11 @@ impl Nac3 {
             tuple: get_attr_id(builtins_mod, "tuple"),
             exception: get_attr_id(builtins_mod, "Exception"),
             option: get_id(artiq_builtins.get_item("Option").ok().flatten().unwrap()),
-            module: get_attr_id(types_mod, "ModuleType"),
         };
 
         let working_directory = tempfile::Builder::new().prefix("nac3-").tempdir().unwrap();
         fs::write(working_directory.path().join("kernel.ld"), include_bytes!("kernel.ld")).unwrap();
 
-        let mut string_store: HashMap<String, i32> = HashMap::default();
-
-        // Keep this list of exceptions in sync with `EXCEPTION_ID_LOOKUP` in `artiq::firmware::ksupport::eh_artiq`
-        // The exceptions declared here must be defined in `artiq.coredevice.exceptions`
-        // Verify synchronization by running the test cases in `artiq.test.coredevice.test_exceptions`
-        let runtime_exception_names = [
-            "RTIOUnderflow",
-            "RTIOOverflow",
-            "RTIODestinationUnreachable",
-            "DMAError",
-            "I2CError",
-            "CacheError",
-            "SPIError",
-            "SubkernelError",
-            "0:AssertionError",
-            "0:AttributeError",
-            "0:IndexError",
-            "0:IOError",
-            "0:KeyError",
-            "0:NotImplementedError",
-            "0:OverflowError",
-            "0:RuntimeError",
-            "0:TimeoutError",
-            "0:TypeError",
-            "0:ValueError",
-            "0:ZeroDivisionError",
-            "0:LinAlgError",
-            "UnwrapNoneError",
-        ];
-
-        // Preallocate runtime exception names
-        for (i, name) in runtime_exception_names.iter().enumerate() {
-            let exn_name = if name.find(':').is_none() {
-                format!("0:artiq.coredevice.exceptions.{name}")
-            } else {
-                (*name).to_string()
-            };
-
-            let id = i32::try_from(i).unwrap();
-            string_store.insert(exn_name, id);
-        }
-
         Ok(Nac3 {
             isa,
             time_fns,
@@ -1189,22 +947,17 @@ impl Nac3 {
             top_levels: Vec::default(),
             pyid_to_def: Arc::default(),
             working_directory,
-            string_store: Arc::new(string_store.into()),
+            string_store: Arc::default(),
             exception_ids: Arc::default(),
             deferred_eval_store: DeferredEvaluationStore::new(),
             llvm_options: CodeGenLLVMOptions {
                 opt_level: OptimizationLevel::Default,
-                target: isa.get_llvm_target_options(),
+                target: Nac3::get_llvm_target_options(isa),
             },
         })
     }
 
-    fn analyze(
-        &mut self,
-        functions: &PySet,
-        classes: &PySet,
-        content_modules: &PySet,
-    ) -> PyResult<()> {
+    fn analyze(&mut self, functions: &PySet, classes: &PySet) -> PyResult<()> {
         let (modules, class_ids) =
             Python::with_gil(|py| -> PyResult<(HashMap<u64, PyObject>, HashSet<u64>)> {
                 let mut modules: HashMap<u64, PyObject> = HashMap::new();
@@ -1214,21 +967,13 @@ impl Nac3 {
                 let getmodule_fn = PyModule::import(py, "inspect")?.getattr("getmodule")?;
 
                 for function in functions {
-                    let module: PyObject = getmodule_fn.call1((function,))?.extract()?;
-                    if !module.is_none(py) {
-                        modules.insert(id_fn.call1((&module,))?.extract()?, module);
-                    }
+                    let module = getmodule_fn.call1((function,))?.extract()?;
+                    modules.insert(id_fn.call1((&module,))?.extract()?, module);
                 }
                 for class in classes {
-                    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()?;
+                    let module = getmodule_fn.call1((class,))?.extract()?;
                     modules.insert(id_fn.call1((&module,))?.extract()?, module);
+                    class_ids.insert(id_fn.call1((class,))?.extract()?);
                 }
                 Ok((modules, class_ids))
             })?;

nac3artiq/src/symbol_resolver.rs

@@ -1,32 +1,14 @@
-use std::{
-    collections::{HashMap, HashSet},
-    sync::{
-        atomic::{AtomicBool, Ordering::Relaxed},
-        Arc,
-    },
+use inkwell::{
+    types::{BasicType, BasicTypeEnum},
+    values::BasicValueEnum,
+    AddressSpace,
 };
-
 use itertools::Itertools;
-use parking_lot::RwLock;
-use pyo3::{
-    types::{PyDict, PyTuple},
-    PyAny, PyErr, PyObject, PyResult, Python,
-};
-
-use super::PrimitivePythonId;
 use nac3core::{
     codegen::{
-        types::{ndarray::NDArrayType, ProxyType},
-        values::ndarray::make_contiguous_strides,
+        classes::{NDArrayType, ProxyType},
         CodeGenContext, CodeGenerator,
     },
-    inkwell::{
-        module::Linkage,
-        types::{BasicType, BasicTypeEnum},
-        values::{BasicValue, BasicValueEnum},
-        AddressSpace,
-    },
-    nac3parser::ast::{self, StrRef},
     symbol_resolver::{StaticValue, SymbolResolver, SymbolValue, ValueEnum},
     toplevel::{
         helper::PrimDef,
@@ -38,6 +20,21 @@ 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;
 
 pub enum PrimitiveValue {
     I32(i32),
@@ -82,6 +79,7 @@ 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,
@@ -135,8 +133,6 @@ 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,
@@ -167,7 +163,7 @@ impl StaticValue for PythonValue {
                 PrimitiveValue::Bool(val) => {
                     ctx.ctx.i8_type().const_int(u64::from(*val), false).into()
                 }
-                PrimitiveValue::Str(val) => ctx.gen_string(generator, val).into(),
+                PrimitiveValue::Str(val) => ctx.ctx.const_string(val.as_bytes(), true).into(),
             });
         }
         if let Some(global) = ctx.module.get_global(&self.id.to_string()) {
@@ -355,7 +351,7 @@ 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![], is_vararg_ctx: false }), false)))
+            Ok(Ok((unifier.add_ty(TypeEnum::TTuple { ty: vec![] }), false)))
         } else if ty_id == self.primitive_ids.option {
             Ok(Ok((primitives.option, false)))
         } else if ty_id == self.primitive_ids.none {
@@ -559,10 +555,7 @@ 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, is_vararg_ctx: false }),
-                        true,
-                    )))
+                    Ok(Ok((unifier.add_ty(TypeEnum::TTuple { ty: args }), true)))
                 }
                 TypeEnum::TObj { params, obj_id, .. } => {
                     let subst = {
@@ -674,48 +667,6 @@ impl InnerResolver {
             })
         });
 
-        // check if obj is module
-        if self.helper.id_fn.call1(py, (ty.clone(),))?.extract::<u64>(py)?
-            == self.primitive_ids.module
-            && self.pyid_to_def.read().contains_key(&py_obj_id)
-        {
-            let def_id = self.pyid_to_def.read()[&py_obj_id];
-            let def = defs[def_id.0].read();
-            let TopLevelDef::Module { name: module_name, module_id, attributes, methods, .. } =
-                &*def
-            else {
-                unreachable!("must be a module here");
-            };
-            // Construct the module return type
-            let mut module_attributes = HashMap::new();
-            for (name, _) in attributes {
-                let attribute_obj = obj.getattr(name.to_string().as_str())?;
-                let attribute_ty =
-                    self.get_obj_type(py, attribute_obj, unifier, defs, primitives)?;
-                if let Ok(attribute_ty) = attribute_ty {
-                    module_attributes.insert(*name, (attribute_ty, false));
-                } else {
-                    return Ok(Err(format!("Unable to resolve {module_name}.{name}")));
-                }
-            }
-
-            for name in methods.keys() {
-                let method_obj = obj.getattr(name.to_string().as_str())?;
-                let method_ty = self.get_obj_type(py, method_obj, unifier, defs, primitives)?;
-                if let Ok(method_ty) = method_ty {
-                    module_attributes.insert(*name, (method_ty, true));
-                } else {
-                    return Ok(Err(format!("Unable to resolve {module_name}.{name}")));
-                }
-            }
-
-            let module_ty =
-                TypeEnum::TModule { module_id: *module_id, attributes: module_attributes };
-
-            let ty = unifier.add_ty(module_ty);
-            return Ok(Ok(ty));
-        }
-
         if let Some(ty) = constructor_ty {
             self.pyid_to_type.write().insert(py_obj_id, ty);
             return Ok(Ok(ty));
@@ -846,9 +797,7 @@ 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, is_vararg_ctx: false })
-                }))
+                Ok(types.map(|types| unifier.add_ty(TypeEnum::TTuple { ty: types })))
             }
             // 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
@@ -973,13 +922,10 @@ impl InnerResolver {
                         |_| Ok(Ok(extracted_ty)),
                     )
                 } else if unifier.unioned(extracted_ty, primitives.bool) {
-                    if obj.extract::<bool>().is_ok()
-                        || obj.call_method("__bool__", (), None)?.extract::<bool>().is_ok()
-                    {
-                        Ok(Ok(extracted_ty))
-                    } else {
-                        Ok(Err(format!("{obj} is not in the range of bool")))
-                    }
+                    obj.extract::<bool>().map_or_else(
+                        |_| Ok(Err(format!("{obj} is not in the range of bool"))),
+                        |_| Ok(Ok(extracted_ty)),
+                    )
                 } else if unifier.unioned(extracted_ty, primitives.float) {
                     obj.extract::<f64>().map_or_else(
                         |_| Ok(Err(format!("{obj} is not in the range of float64"))),
@@ -1019,18 +965,14 @@ impl InnerResolver {
             let val: u64 = obj.extract().unwrap();
             self.id_to_primitive.write().insert(id, PrimitiveValue::U64(val));
             Ok(Some(ctx.ctx.i64_type().const_int(val, false).into()))
-        } else if ty_id == self.primitive_ids.bool {
+        } else if ty_id == self.primitive_ids.bool || ty_id == self.primitive_ids.np_bool_ {
             let val: bool = obj.extract().unwrap();
             self.id_to_primitive.write().insert(id, PrimitiveValue::Bool(val));
             Ok(Some(ctx.ctx.i8_type().const_int(u64::from(val), false).into()))
-        } else if ty_id == self.primitive_ids.np_bool_ {
-            let val: bool = obj.call_method("__bool__", (), None)?.extract().unwrap();
-            self.id_to_primitive.write().insert(id, PrimitiveValue::Bool(val));
-            Ok(Some(ctx.ctx.i8_type().const_int(u64::from(val), false).into()))
         } 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.gen_string(generator, val).into()))
+            Ok(Some(ctx.ctx.const_string(val.as_bytes(), true).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));
@@ -1049,15 +991,8 @@ impl InnerResolver {
                 }
                 _ => unreachable!("must be list"),
             };
-            let size_t = ctx.get_size_type();
-            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 ty = ctx.get_llvm_type(generator, elem_ty);
+            let size_t = generator.get_size_type(ctx.ctx);
             let arr_ty = ctx
                 .ctx
                 .struct_type(&[ty.ptr_type(AddressSpace::default()).into(), size_t.into()], false);
@@ -1134,19 +1069,18 @@ impl InnerResolver {
             } else {
                 unreachable!("must be ndarray")
             };
-            let (ndarray_dtype, _) = unpack_ndarray_var_tys(&mut ctx.unifier, ndarray_ty);
+            let (ndarray_dtype, ndarray_ndims) =
+                unpack_ndarray_var_tys(&mut ctx.unifier, ndarray_ty);
 
-            let llvm_i8 = ctx.ctx.i8_type();
-            let llvm_pi8 = llvm_i8.ptr_type(AddressSpace::default());
-            let llvm_usize = ctx.get_size_type();
-            let llvm_ndarray = NDArrayType::from_unifier_type(generator, ctx, ndarray_ty);
-            let dtype = llvm_ndarray.element_type();
+            let llvm_usize = generator.get_size_type(ctx.ctx);
+            let ndarray_dtype_llvm_ty = ctx.get_llvm_type(generator, ndarray_dtype);
+            let ndarray_llvm_ty = NDArrayType::new(generator, ctx.ctx, ndarray_dtype_llvm_ty);
 
             {
                 if self.global_value_ids.read().contains_key(&id) {
                     let global = ctx.module.get_global(&id_str).unwrap_or_else(|| {
                         ctx.module.add_global(
-                            llvm_ndarray.as_base_type().get_element_type().into_struct_type(),
+                            ndarray_llvm_ty.as_underlying_type(),
                             Some(AddressSpace::default()),
                             &id_str,
                         )
@@ -1156,44 +1090,40 @@ impl InnerResolver {
                 self.global_value_ids.write().insert(id, obj.into());
             }
 
-            let ndims = llvm_ndarray.ndims();
+            let TypeEnum::TLiteral { values, .. } = &*ctx.unifier.get_ty_immutable(ndarray_ndims)
+            else {
+                unreachable!("Expected Literal for ndarray_ndims")
+            };
+
+            let ndarray_ndims = if values.len() == 1 {
+                values[0].clone()
+            } else {
+                todo!("Unpacking literal of more than one element unimplemented")
+            };
+            let Ok(ndarray_ndims) = u64::try_from(ndarray_ndims) else {
+                unreachable!("Expected u64 value for ndarray_ndims")
+            };
 
             // Obtain the shape of the ndarray
             let shape_tuple: &PyTuple = obj.getattr("shape")?.downcast()?;
-            assert_eq!(shape_tuple.len(), ndims as usize);
-
-            // The Rust type inferencer cannot figure this out
-            let shape_values = shape_tuple
+            assert_eq!(shape_tuple.len(), ndarray_ndims as usize);
+            let shape_values: Result<Option<Vec<_>>, _> = shape_tuple
                 .iter()
                 .enumerate()
                 .map(|(i, elem)| {
-                    let value = self
-                        .get_obj_value(py, elem, ctx, generator, ctx.primitives.usize())
-                        .map_err(|e| {
-                            super::CompileError::new_err(format!("Error getting element {i}: {e}"))
-                        })?
-                        .unwrap();
-                    let value = ctx
-                        .builder
-                        .build_int_z_extend(value.into_int_value(), llvm_usize, "")
-                        .unwrap();
-                    Ok(value)
+                    self.get_obj_value(py, elem, ctx, generator, ctx.primitives.usize()).map_err(
+                        |e| super::CompileError::new_err(format!("Error getting element {i}: {e}")),
+                    )
                 })
-                .collect::<Result<Vec<_>, PyErr>>()?;
-
-            // Also use this opportunity to get the constant values of `shape_values` for calculating strides.
-            let shape_u64s = shape_values
-                .iter()
-                .map(|dim| {
-                    assert!(dim.is_const());
-                    dim.get_zero_extended_constant().unwrap()
-                })
-                .collect_vec();
-            let shape_values = llvm_usize.const_array(&shape_values);
+                .collect();
+            let shape_values = shape_values?.unwrap();
+            let shape_values = llvm_usize.const_array(
+                &shape_values.into_iter().map(BasicValueEnum::into_int_value).collect_vec(),
+            );
 
             // create a global for ndarray.shape and initialize it using the shape
             let shape_global = ctx.module.add_global(
-                llvm_usize.array_type(ndims as u32),
+                llvm_usize.array_type(ndarray_ndims as u32),
                 Some(AddressSpace::default()),
                 &(id_str.clone() + ".shape"),
             );
@@ -1201,25 +1131,17 @@ impl InnerResolver {
 
             // Obtain the (flattened) elements of the ndarray
             let sz: usize = obj.getattr("size")?.extract()?;
-            let data: Vec<_> = (0..sz)
+            let data: Result<Option<Vec<_>>, _> = (0..sz)
                 .map(|i| {
                     obj.getattr("flat")?.get_item(i).and_then(|elem| {
-                        let value = self
-                            .get_obj_value(py, elem, ctx, generator, ndarray_dtype)
-                            .map_err(|e| {
-                                super::CompileError::new_err(format!(
-                                    "Error getting element {i}: {e}"
-                                ))
-                            })?
-                            .unwrap();
-
-                        assert_eq!(value.get_type(), dtype);
-                        Ok(value)
+                        self.get_obj_value(py, elem, ctx, generator, ndarray_dtype).map_err(|e| {
+                            super::CompileError::new_err(format!("Error getting element {i}: {e}"))
+                        })
                     })
                 })
-                .try_collect()?;
-            let data = data.into_iter();
-            let data = match dtype {
+                .collect();
+            let data = data?.unwrap().into_iter();
+            let data = match ndarray_dtype_llvm_ty {
                 BasicTypeEnum::ArrayType(ty) => {
                     ty.const_array(&data.map(BasicValueEnum::into_array_value).collect_vec())
                 }
@@ -1244,102 +1166,37 @@ impl InnerResolver {
             };
 
             // create a global for ndarray.data and initialize it using the elements
-            //
-            // NOTE: NDArray's `data` is `u8*`. Here, `data_global` is an array of `dtype`.
-            // We will have to cast it to an `u8*` later.
             let data_global = ctx.module.add_global(
-                dtype.array_type(sz as u32),
+                ndarray_dtype_llvm_ty.array_type(sz as u32),
                 Some(AddressSpace::default()),
                 &(id_str.clone() + ".data"),
             );
             data_global.set_initializer(&data);
 
-            // Get the constant itemsize.
-            //
-            // NOTE: dtype.size_of() may return a non-constant, where `TargetData::get_store_size`
-            // will always return a constant size.
-            let itemsize = ctx
-                .registry
-                .llvm_options
-                .create_target_machine()
-                .map(|tm| tm.get_target_data().get_store_size(&dtype))
-                .unwrap();
-            assert_ne!(itemsize, 0);
-
-            // Create the strides needed for ndarray.strides
-            let strides = make_contiguous_strides(itemsize, ndims, &shape_u64s);
-            let strides =
-                strides.into_iter().map(|stride| llvm_usize.const_int(stride, false)).collect_vec();
-            let strides = llvm_usize.const_array(&strides);
-
-            // create a global for ndarray.strides and initialize it
-            let strides_global = ctx.module.add_global(
-                llvm_usize.array_type(ndims as u32),
-                Some(AddressSpace::default()),
-                &format!("${id_str}.strides"),
-            );
-            strides_global.set_initializer(&strides);
-
             // create a global for the ndarray object and initialize it
+            let value = ndarray_llvm_ty.as_underlying_type().const_named_struct(&[
+                llvm_usize.const_int(ndarray_ndims, false).into(),
+                shape_global
+                    .as_pointer_value()
+                    .const_cast(llvm_usize.ptr_type(AddressSpace::default()))
+                    .into(),
+                data_global
+                    .as_pointer_value()
+                    .const_cast(ndarray_dtype_llvm_ty.ptr_type(AddressSpace::default()))
+                    .into(),
+            ]);
 
-            // NOTE: data_global is an array of dtype, we want a `u8*`.
-            let ndarray_data = data_global.as_pointer_value();
-            let ndarray_data = ctx.builder.build_pointer_cast(ndarray_data, llvm_pi8, "").unwrap();
-
-            let ndarray_itemsize = llvm_usize.const_int(itemsize, false);
-
-            let ndarray_ndims = llvm_usize.const_int(ndims, false);
-
-            // calling as_pointer_value on shape and strides returns [i64 x ndims]*
-            // convert into i64* to conform with expected layout of ndarray
-
-            let ndarray_shape = shape_global.as_pointer_value();
-            let ndarray_shape = unsafe {
-                ctx.builder
-                    .build_in_bounds_gep(
-                        ndarray_shape,
-                        &[llvm_usize.const_zero(), llvm_usize.const_zero()],
-                        "",
-                    )
-                    .unwrap()
-            };
-
-            let ndarray_strides = strides_global.as_pointer_value();
-            let ndarray_strides = unsafe {
-                ctx.builder
-                    .build_in_bounds_gep(
-                        ndarray_strides,
-                        &[llvm_usize.const_zero(), llvm_usize.const_zero()],
-                        "",
-                    )
-                    .unwrap()
-            };
-
-            let ndarray = llvm_ndarray
-                .as_base_type()
-                .get_element_type()
-                .into_struct_type()
-                .const_named_struct(&[
-                    ndarray_itemsize.into(),
-                    ndarray_ndims.into(),
-                    ndarray_shape.into(),
-                    ndarray_strides.into(),
-                    ndarray_data.into(),
-                ]);
-
-            let ndarray_global = ctx.module.add_global(
-                llvm_ndarray.as_base_type().get_element_type().into_struct_type(),
+            let ndarray = ctx.module.add_global(
+                ndarray_llvm_ty.as_underlying_type(),
                 Some(AddressSpace::default()),
                 &id_str,
             );
-            ndarray_global.set_initializer(&ndarray);
+            ndarray.set_initializer(&value);
 
-            Ok(Some(ndarray_global.as_pointer_value().into()))
+            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, is_vararg_ctx: false } = expected_ty_enum.as_ref() else {
-                unreachable!()
-            };
+            let TypeEnum::TTuple { ty } = expected_ty_enum.as_ref() else { unreachable!() };
 
             let tup_tys = ty.iter();
             let elements: &PyTuple = obj.downcast()?;
@@ -1415,77 +1272,6 @@ impl InnerResolver {
                     None => Ok(None),
                 }
             }
-        } else if ty_id == self.primitive_ids.module {
-            let id_str = id.to_string();
-
-            if let Some(global) = ctx.module.get_global(&id_str) {
-                return Ok(Some(global.as_pointer_value().into()));
-            }
-
-            let top_level_defs = ctx.top_level.definitions.read();
-            let ty = self
-                .get_obj_type(py, obj, &mut ctx.unifier, &top_level_defs, &ctx.primitives)?
-                .unwrap();
-            let ty = ctx
-                .get_llvm_type(generator, ty)
-                .into_pointer_type()
-                .get_element_type()
-                .into_struct_type();
-
-            {
-                if self.global_value_ids.read().contains_key(&id) {
-                    let global = ctx.module.get_global(&id_str).unwrap_or_else(|| {
-                        ctx.module.add_global(ty, Some(AddressSpace::default()), &id_str)
-                    });
-                    return Ok(Some(global.as_pointer_value().into()));
-                }
-                self.global_value_ids.write().insert(id, obj.into());
-            }
-
-            let fields = {
-                let definition =
-                    top_level_defs.get(self.pyid_to_def.read().get(&id).unwrap().0).unwrap().read();
-                let TopLevelDef::Module { attributes, .. } = &*definition else { unreachable!() };
-                attributes
-                    .iter()
-                    .filter_map(|f| {
-                        let definition = top_level_defs.get(f.1 .0).unwrap().read();
-                        if let TopLevelDef::Variable { ty, .. } = &*definition {
-                            Some((f.0, *ty))
-                        } else {
-                            None
-                        }
-                    })
-                    .collect_vec()
-            };
-
-            let values: Result<Option<Vec<_>>, _> = fields
-                .iter()
-                .map(|(name, ty)| {
-                    self.get_obj_value(
-                        py,
-                        obj.getattr(name.to_string().as_str())?,
-                        ctx,
-                        generator,
-                        *ty,
-                    )
-                    .map_err(|e| {
-                        super::CompileError::new_err(format!("Error getting field {name}: {e}"))
-                    })
-                })
-                .collect();
-            let values = values?;
-
-            if let Some(values) = values {
-                let val = ty.const_named_struct(&values);
-                let global = ctx.module.get_global(&id_str).unwrap_or_else(|| {
-                    ctx.module.add_global(ty, Some(AddressSpace::default()), &id_str)
-                });
-                global.set_initializer(&val);
-                Ok(Some(global.as_pointer_value().into()))
-            } else {
-                Ok(None)
-            }
         } else {
             let id_str = id.to_string();
 
@@ -1565,12 +1351,9 @@ impl InnerResolver {
         } else if ty_id == self.primitive_ids.uint64 {
             let val: u64 = obj.extract()?;
             Ok(SymbolValue::U64(val))
-        } else if ty_id == self.primitive_ids.bool {
+        } else if ty_id == self.primitive_ids.bool || ty_id == self.primitive_ids.np_bool_ {
             let val: bool = obj.extract()?;
             Ok(SymbolValue::Bool(val))
-        } else if ty_id == self.primitive_ids.np_bool_ {
-            let val: bool = obj.call_method("__bool__", (), None)?.extract()?;
-            Ok(SymbolValue::Bool(val))
         } else if ty_id == self.primitive_ids.string || ty_id == self.primitive_ids.np_str_ {
             let val: String = obj.extract()?;
             Ok(SymbolValue::Str(val))
@@ -1668,50 +1451,8 @@ impl SymbolResolver for Resolver {
     fn get_symbol_value<'ctx>(
         &self,
         id: StrRef,
-        ctx: &mut CodeGenContext<'ctx, '_>,
-        generator: &mut dyn CodeGenerator,
+        _: &mut CodeGenContext<'ctx, '_>,
     ) -> Option<ValueEnum<'ctx>> {
-        if let Some(def_id) = self.0.id_to_def.read().get(&id) {
-            let top_levels = ctx.top_level.definitions.read();
-            if matches!(&*top_levels[def_id.0].read(), TopLevelDef::Variable { .. }) {
-                let module_val = &self.0.module;
-                let ret = Python::with_gil(|py| -> PyResult<Result<BasicValueEnum, String>> {
-                    let module_val = module_val.as_ref(py);
-
-                    let ty = self.0.get_obj_type(
-                        py,
-                        module_val,
-                        &mut ctx.unifier,
-                        &top_levels,
-                        &ctx.primitives,
-                    )?;
-                    if let Err(ty) = ty {
-                        return Ok(Err(ty));
-                    }
-                    let ty = ty.unwrap();
-                    let obj = self.0.get_obj_value(py, module_val, ctx, generator, ty)?.unwrap();
-                    let (idx, _) = ctx.get_attr_index(ty, id);
-                    let ret = unsafe {
-                        ctx.builder.build_gep(
-                            obj.into_pointer_value(),
-                            &[
-                                ctx.ctx.i32_type().const_zero(),
-                                ctx.ctx.i32_type().const_int(idx as u64, false),
-                            ],
-                            id.to_string().as_str(),
-                        )
-                    }
-                    .unwrap();
-                    Ok(Ok(ret.as_basic_value_enum()))
-                })
-                .unwrap();
-                if ret.is_err() {
-                    return None;
-                }
-                return Some(ret.unwrap().into());
-            }
-        }
-
         let sym_value = {
             let id_to_val = self.0.id_to_pyval.read();
             id_to_val.get(&id).cloned()
@@ -1772,7 +1513,10 @@ impl SymbolResolver for Resolver {
         if let Some(id) = string_store.get(s) {
             *id
         } else {
-            let id = i32::try_from(string_store.len()).unwrap();
+            let id = Python::with_gil(|py| -> PyResult<i32> {
+                self.0.helper.store_str.call1(py, (s,))?.extract(py)
+            })
+            .unwrap();
             string_store.insert(s.into(), id);
             id
         }

nac3artiq/src/timeline.rs

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

nac3ast/Cargo.toml

@@ -10,6 +10,7 @@ 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

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

nac3ast/src/lib.rs

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

nac3core/Cargo.toml

@@ -1,29 +1,26 @@
+[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 = "2.6"
+indexmap = "2.2"
 parking_lot = "0.12"
-rayon = "1.10"
-nac3core_derive = { path = "nac3core_derive", optional = true }
+rayon = "1.8"
 nac3parser = { path = "../nac3parser" }
-strum = "0.26"
-strum_macros = "0.26"
+strum = "0.26.2"
+strum_macros = "0.26.4"
 
 [dependencies.inkwell]
-version = "0.5"
+version = "0.4"
 default-features = false
-features = ["llvm14-0-prefer-dynamic", "target-x86", "target-arm", "target-riscv", "no-libffi-linking"]
+features = ["llvm14-0", "target-x86", "target-arm", "target-riscv", "no-libffi-linking"]
 
 [dev-dependencies]
 test-case = "1.2.0"

nac3core/build.rs

@@ -1,63 +1,68 @@
+use regex::Regex;
 use std::{
     env,
     fs::File,
     io::Write,
-    path::Path,
+    path::{Path, PathBuf},
     process::{Command, Stdio},
 };
 
-use regex::Regex;
+const CMD_IRRT_CLANG: &str = "clang-irrt";
+const CMD_IRRT_CLANG_TEST: &str = "clang-irrt-test";
+const CMD_IRRT_LLVM_AS: &str = "llvm-as-irrt";
 
-fn main() {
-    let out_dir = env::var("OUT_DIR").unwrap();
-    let out_dir = Path::new(&out_dir);
-    let irrt_dir = Path::new("irrt");
+fn get_out_dir() -> PathBuf {
+    PathBuf::from(env::var("OUT_DIR").unwrap())
+}
 
-    let irrt_cpp_path = irrt_dir.join("irrt.cpp");
+fn get_irrt_dir() -> &'static Path {
+    Path::new("irrt")
+}
+
+/// Compile `irrt.cpp` for use in `src/codegen`
+fn compile_irrt_cpp() {
+    let out_dir = get_out_dir();
+    let irrt_dir = get_irrt_dir();
 
     /*
      * 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 mut flags: Vec<&str> = vec![
+    let irrt_cpp_path = irrt_dir.join("irrt.cpp");
+    let flags: &[&str] = &[
         "--target=wasm32",
         "-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",
-        "-o",
-        "-",
+        "-Werror=return-type",
         "-I",
         irrt_dir.to_str().unwrap(),
+        "-o",
+        "-",
         irrt_cpp_path.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")
+    let output = Command::new(CMD_IRRT_CLANG)
         .args(flags)
         .output()
-        .inspect(|o| {
+        .map(|o| {
             assert!(o.status.success(), "{}", std::str::from_utf8(&o.stderr).unwrap());
+            o
         })
         .unwrap();
 
@@ -98,7 +103,9 @@ fn main() {
         file.write_all(filtered_output.as_bytes()).unwrap();
     }
 
-    let mut llvm_as = Command::new("llvm-as-irrt")
+    // Assemble the emitted and filtered IR to .bc
+    // That .bc will be integrated into nac3core's codegen
+    let mut llvm_as = Command::new(CMD_IRRT_LLVM_AS)
         .stdin(Stdio::piped())
         .arg("-o")
         .arg(out_dir.join("irrt.bc"))
@@ -107,3 +114,50 @@ fn main() {
     llvm_as.stdin.as_mut().unwrap().write_all(filtered_output.as_bytes()).unwrap();
     assert!(llvm_as.wait().unwrap().success());
 }
+
+/// Compile `irrt_test.cpp` for testing
+fn compile_irrt_test_cpp() {
+    let out_dir = get_out_dir();
+    let irrt_dir = get_irrt_dir();
+
+    let exe_path = out_dir.join("irrt_test.out"); // Output path of the compiled test executable
+    let irrt_test_cpp_path = irrt_dir.join("irrt_test.cpp");
+    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()`
+        "-I",
+        irrt_dir.to_str().unwrap(),
+        "-o",
+        exe_path.to_str().unwrap(),
+    ];
+
+    Command::new(CMD_IRRT_CLANG_TEST)
+        .args(flags)
+        .output()
+        .map(|o| {
+            assert!(o.status.success(), "{}", std::str::from_utf8(&o.stderr).unwrap());
+            o
+        })
+        .unwrap();
+    println!("cargo:rerun-if-changed={}", irrt_dir.to_str().unwrap());
+}
+
+fn main() {
+    compile_irrt_cpp();
+
+    // 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_cpp();
+    }
+}

nac3core/irrt/irrt.cpp

@@ -1,15 +1,9 @@
-#include "irrt/exception.hpp"
-#include "irrt/list.hpp"
-#include "irrt/math.hpp"
-#include "irrt/range.hpp"
-#include "irrt/slice.hpp"
-#include "irrt/string.hpp"
-#include "irrt/ndarray/basic.hpp"
-#include "irrt/ndarray/def.hpp"
-#include "irrt/ndarray/iter.hpp"
-#include "irrt/ndarray/indexing.hpp"
-#include "irrt/ndarray/array.hpp"
-#include "irrt/ndarray/reshape.hpp"
-#include "irrt/ndarray/broadcast.hpp"
-#include "irrt/ndarray/transpose.hpp"
-#include "irrt/ndarray/matmul.hpp"
+#define IRRT_DEFINE_TYPEDEF_INTS
+#include <irrt_everything.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/core.hpp

@@ -0,0 +1,402 @@
+#pragma once
+
+#include <irrt/utils.hpp>
+#include <irrt/int_defs.hpp>
+
+// 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;
+
+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;
+}
+
+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" {
+#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);
+}
+}  // extern "C"

nac3core/irrt/irrt/cslice.hpp

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

nac3core/irrt/irrt/debug.hpp

@@ -1,25 +0,0 @@
-#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/error_context.hpp

@@ -0,0 +1,85 @@
+#pragma once
+
+#include <irrt/int_defs.hpp>
+#include <irrt/utils.hpp>
+
+namespace {
+    // nac3core's "str" struct type definition
+    template <typename SizeT>
+    struct Str {
+        const char* content;
+        SizeT length;
+    };
+
+    // A limited set of errors IRRT could use.
+    typedef uint32_t ErrorId;
+    struct ErrorIds {
+        ErrorId index_error;
+        ErrorId value_error;
+        ErrorId assertion_error;
+        ErrorId runtime_error;
+        ErrorId type_error;
+    };
+
+    struct ErrorContext {
+        // Context
+        const ErrorIds* error_ids;
+
+        // Error thrown by IRRT
+        ErrorId error_id;
+        const char* message_template; // MUST BE `&'static`
+        int64_t param1;
+        int64_t param2;
+        int64_t param3;
+
+        void initialize(const ErrorIds* error_ids) {
+            this->error_ids = error_ids;
+            clear_error();
+        }
+
+        void clear_error() {
+            // Point the message_template to an empty str. Don't set it to nullptr as a sentinel
+            this->message_template = "";
+        }
+
+        void set_error(ErrorId error_id, const char* message, int64_t param1 = 0, int64_t param2 = 0, int64_t param3 = 0) {
+            this->error_id = error_id;
+            this->message_template = message;
+            this->param1 = param1;
+            this->param2 = param2;
+            this->param3 = param3;
+        }
+
+        bool has_error() {
+            return !cstr_utils::is_empty(message_template);
+        }
+
+        template <typename SizeT>
+        void get_error_str(Str<SizeT> *dst_str) {
+            dst_str->content = message_template;
+            dst_str->length = (SizeT) cstr_utils::length(message_template);
+        }
+    };
+}
+
+extern "C" {
+    void __nac3_error_context_initialize(ErrorContext* errctx, const ErrorIds* error_ids) {
+        errctx->initialize(error_ids);
+    }
+
+    bool __nac3_error_context_has_no_error(ErrorContext* errctx) {
+        return !errctx->has_error();
+    }
+
+    void __nac3_error_context_get_error_str(ErrorContext* errctx, Str<int32_t> *dst_str) {
+        errctx->get_error_str<int32_t>(dst_str);
+    }
+
+    void __nac3_error_context_get_error_str64(ErrorContext* errctx, Str<int64_t> *dst_str) {
+        errctx->get_error_str<int64_t>(dst_str);
+    }
+
+    void __nac3_error_dummy_raise(ErrorContext* errctx) {
+        errctx->set_error(errctx->error_ids->runtime_error, "THROWN FROM __nac3_error_dummy_raise!!!!!!");
+    }
+}

nac3core/irrt/irrt/exception.hpp

@@ -1,85 +0,0 @@
-#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_defs.hpp

@@ -0,0 +1,12 @@
+#pragma once
+
+// This is made toggleable since `irrt_test.cpp` itself would include
+// headers that define these typedefs
+#ifdef IRRT_DEFINE_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

nac3core/irrt/irrt/int_types.hpp

@@ -1,25 +0,0 @@
-#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
-
-// 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

@@ -1,96 +0,0 @@
-#pragma once
-
-#include "irrt/int_types.hpp"
-#include "irrt/math_util.hpp"
-#include "irrt/slice.hpp"
-
-namespace {
-/**
- * @brief A list in NAC3.
- *
- * The `items` field is opaque. You must rely on external contexts to
- * know how to interpret it.
- */
-template<typename SizeT>
-struct List {
-    uint8_t* items;
-    SizeT len;
-};
-}  // namespace
-
-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

@@ -1,95 +0,0 @@
-#pragma once
-
-#include "irrt/int_types.hpp"
-
-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

@@ -1,13 +0,0 @@
-#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/array.hpp

@@ -1,132 +0,0 @@
-#pragma once
-
-#include "irrt/debug.hpp"
-#include "irrt/exception.hpp"
-#include "irrt/int_types.hpp"
-#include "irrt/list.hpp"
-#include "irrt/ndarray/basic.hpp"
-#include "irrt/ndarray/def.hpp"
-
-namespace {
-namespace ndarray::array {
-/**
- * @brief In the context of `np.array(<list>)`, deduce the ndarray's shape produced by `<list>` and raise
- * an exception if there is anything wrong with `<shape>` (e.g., inconsistent dimensions `np.array([[1.0, 2.0],
- * [3.0]])`)
- *
- * If this function finds no issues with `<list>`, the deduced shape is written to `shape`. The caller has the
- * responsibility to allocate `[SizeT; ndims]` for `shape`. The caller must also initialize `shape` with `-1`s because
- * of implementation details.
- */
-template<typename SizeT>
-void set_and_validate_list_shape_helper(SizeT axis, List<SizeT>* list, SizeT ndims, SizeT* shape) {
-    if (shape[axis] == -1) {
-        // Dimension is unspecified. Set it.
-        shape[axis] = list->len;
-    } else {
-        // Dimension is specified. Check.
-        if (shape[axis] != list->len) {
-            // Mismatch, throw an error.
-            // NOTE: NumPy's error message is more complex and needs more PARAMS to display.
-            raise_exception(SizeT, EXN_VALUE_ERROR,
-                            "The requested array has an inhomogenous shape "
-                            "after {0} dimension(s).",
-                            axis, shape[axis], list->len);
-        }
-    }
-
-    if (axis + 1 == ndims) {
-        // `list` has type `list[ItemType]`
-        // Do nothing
-    } else {
-        // `list` has type `list[list[...]]`
-        List<SizeT>** lists = (List<SizeT>**)(list->items);
-        for (SizeT i = 0; i < list->len; i++) {
-            set_and_validate_list_shape_helper<SizeT>(axis + 1, lists[i], ndims, shape);
-        }
-    }
-}
-
-/**
- * @brief See `set_and_validate_list_shape_helper`.
- */
-template<typename SizeT>
-void set_and_validate_list_shape(List<SizeT>* list, SizeT ndims, SizeT* shape) {
-    for (SizeT axis = 0; axis < ndims; axis++) {
-        shape[axis] = -1;  // Sentinel to say this dimension is unspecified.
-    }
-    set_and_validate_list_shape_helper<SizeT>(0, list, ndims, shape);
-}
-
-/**
- * @brief In the context of `np.array(<list>)`, copied the contents stored in `list` to `ndarray`.
- *
- * `list` is assumed to be "legal". (i.e., no inconsistent dimensions)
- *
- * # Notes on `ndarray`
- * The caller is responsible for allocating space for `ndarray`.
- * Here is what this function expects from `ndarray` when called:
- *   - `ndarray->data` has to be allocated, contiguous, and may contain uninitialized values.
- *   - `ndarray->itemsize` has to be initialized.
- *   - `ndarray->ndims` has to be initialized.
- *   - `ndarray->shape` has to be initialized.
- *   - `ndarray->strides` is ignored, but note that `ndarray->data` is contiguous.
- * When this function call ends:
- *   - `ndarray->data` is written with contents from `<list>`.
- */
-template<typename SizeT>
-void write_list_to_array_helper(SizeT axis, SizeT* index, List<SizeT>* list, NDArray<SizeT>* ndarray) {
-    debug_assert_eq(SizeT, list->len, ndarray->shape[axis]);
-    if (IRRT_DEBUG_ASSERT_BOOL) {
-        if (!ndarray::basic::is_c_contiguous(ndarray)) {
-            raise_debug_assert(SizeT, "ndarray is not C-contiguous", ndarray->strides[0], ndarray->strides[1],
-                               NO_PARAM);
-        }
-    }
-
-    if (axis + 1 == ndarray->ndims) {
-        // `list` has type `list[scalar]`
-        // `ndarray` is contiguous, so we can do this, and this is fast.
-        uint8_t* dst = static_cast<uint8_t*>(ndarray->data) + (ndarray->itemsize * (*index));
-        __builtin_memcpy(dst, list->items, ndarray->itemsize * list->len);
-        *index += list->len;
-    } else {
-        // `list` has type `list[list[...]]`
-        List<SizeT>** lists = (List<SizeT>**)(list->items);
-
-        for (SizeT i = 0; i < list->len; i++) {
-            write_list_to_array_helper<SizeT>(axis + 1, index, lists[i], ndarray);
-        }
-    }
-}
-
-/**
- * @brief See `write_list_to_array_helper`.
- */
-template<typename SizeT>
-void write_list_to_array(List<SizeT>* list, NDArray<SizeT>* ndarray) {
-    SizeT index = 0;
-    write_list_to_array_helper<SizeT>((SizeT)0, &index, list, ndarray);
-}
-}  // namespace ndarray::array
-}  // namespace
-
-extern "C" {
-using namespace ndarray::array;
-
-void __nac3_ndarray_array_set_and_validate_list_shape(List<int32_t>* list, int32_t ndims, int32_t* shape) {
-    set_and_validate_list_shape(list, ndims, shape);
-}
-
-void __nac3_ndarray_array_set_and_validate_list_shape64(List<int64_t>* list, int64_t ndims, int64_t* shape) {
-    set_and_validate_list_shape(list, ndims, shape);
-}
-
-void __nac3_ndarray_array_write_list_to_array(List<int32_t>* list, NDArray<int32_t>* ndarray) {
-    write_list_to_array(list, ndarray);
-}
-
-void __nac3_ndarray_array_write_list_to_array64(List<int64_t>* list, NDArray<int64_t>* ndarray) {
-    write_list_to_array(list, ndarray);
-}
-}

nac3core/irrt/irrt/ndarray/basic.hpp

@@ -1,340 +0,0 @@
-#pragma once
-
-#include "irrt/debug.hpp"
-#include "irrt/exception.hpp"
-#include "irrt/int_types.hpp"
-#include "irrt/ndarray/def.hpp"
-
-namespace {
-namespace ndarray::basic {
-/**
- * @brief Assert that `shape` does not contain negative dimensions.
- *
- * @param ndims Number of dimensions in `shape`
- * @param shape The shape to check on
- */
-template<typename SizeT>
-void assert_shape_no_negative(SizeT ndims, const SizeT* shape) {
-    for (SizeT axis = 0; axis < ndims; axis++) {
-        if (shape[axis] < 0) {
-            raise_exception(SizeT, EXN_VALUE_ERROR,
-                            "negative dimensions are not allowed; axis {0} "
-                            "has dimension {1}",
-                            axis, shape[axis], NO_PARAM);
-        }
-    }
-}
-
-/**
- * @brief Assert that two shapes are the same in the context of writing output to an ndarray.
- */
-template<typename SizeT>
-void assert_output_shape_same(SizeT ndarray_ndims,
-                              const SizeT* ndarray_shape,
-                              SizeT output_ndims,
-                              const SizeT* output_shape) {
-    if (ndarray_ndims != output_ndims) {
-        // There is no corresponding NumPy error message like this.
-        raise_exception(SizeT, EXN_VALUE_ERROR, "Cannot write output of ndims {0} to an ndarray with ndims {1}",
-                        output_ndims, ndarray_ndims, NO_PARAM);
-    }
-
-    for (SizeT axis = 0; axis < ndarray_ndims; axis++) {
-        if (ndarray_shape[axis] != output_shape[axis]) {
-            // There is no corresponding NumPy error message like this.
-            raise_exception(SizeT, EXN_VALUE_ERROR,
-                            "Mismatched dimensions on axis {0}, output has "
-                            "dimension {1}, but destination ndarray has dimension {2}.",
-                            axis, output_shape[axis], ndarray_shape[axis]);
-        }
-    }
-}
-
-/**
- * @brief Return the number of elements of an ndarray given its shape.
- *
- * @param ndims Number of dimensions in `shape`
- * @param shape The shape of the ndarray
- */
-template<typename SizeT>
-SizeT calc_size_from_shape(SizeT ndims, const SizeT* shape) {
-    SizeT size = 1;
-    for (SizeT axis = 0; axis < ndims; axis++)
-        size *= shape[axis];
-    return size;
-}
-
-/**
- * @brief Compute the array indices of the `nth` (0-based) element of an ndarray given only its shape.
- *
- * @param ndims Number of elements in `shape` and `indices`
- * @param shape The shape of the ndarray
- * @param indices The returned indices indexing the ndarray with shape `shape`.
- * @param nth The index of the element of interest.
- */
-template<typename SizeT>
-void set_indices_by_nth(SizeT ndims, const SizeT* shape, SizeT* indices, SizeT nth) {
-    for (SizeT i = 0; i < ndims; i++) {
-        SizeT axis = ndims - i - 1;
-        SizeT dim = shape[axis];
-
-        indices[axis] = nth % dim;
-        nth /= dim;
-    }
-}
-
-/**
- * @brief Return the number of elements of an `ndarray`
- *
- * This function corresponds to `<an_ndarray>.size`
- */
-template<typename SizeT>
-SizeT size(const NDArray<SizeT>* ndarray) {
-    return calc_size_from_shape(ndarray->ndims, ndarray->shape);
-}
-
-/**
- * @brief Return of the number of its content of an `ndarray`.
- *
- * This function corresponds to `<an_ndarray>.nbytes`.
- */
-template<typename SizeT>
-SizeT nbytes(const NDArray<SizeT>* ndarray) {
-    return size(ndarray) * ndarray->itemsize;
-}
-
-/**
- * @brief Get the `len()` of an ndarray, and asserts that `ndarray` is a sized object.
- *
- * This function corresponds to `<an_ndarray>.__len__`.
- *
- * @param dst_length The length.
- */
-template<typename SizeT>
-SizeT len(const NDArray<SizeT>* ndarray) {
-    if (ndarray->ndims != 0) {
-        return ndarray->shape[0];
-    }
-
-    // numpy prohibits `__len__` on unsized objects
-    raise_exception(SizeT, EXN_TYPE_ERROR, "len() of unsized object", NO_PARAM, NO_PARAM, NO_PARAM);
-    __builtin_unreachable();
-}
-
-/**
- * @brief Return a boolean indicating if `ndarray` is (C-)contiguous.
- *
- * You may want to see ndarray's rules for C-contiguity:
- * https://github.com/numpy/numpy/blob/df256d0d2f3bc6833699529824781c58f9c6e697/numpy/core/src/multiarray/flagsobject.c#L95C1-L99C45
- */
-template<typename SizeT>
-bool is_c_contiguous(const NDArray<SizeT>* ndarray) {
-    // References:
-    // - tinynumpy's implementation:
-    // https://github.com/wadetb/tinynumpy/blob/0d23d22e07062ffab2afa287374c7b366eebdda1/tinynumpy/tinynumpy.py#L102
-    // - ndarray's flags["C_CONTIGUOUS"]:
-    // https://numpy.org/doc/stable/reference/generated/numpy.ndarray.flags.html#numpy.ndarray.flags
-    // - ndarray's rules for C-contiguity:
-    // https://github.com/numpy/numpy/blob/df256d0d2f3bc6833699529824781c58f9c6e697/numpy/core/src/multiarray/flagsobject.c#L95C1-L99C45
-
-    // From
-    // https://github.com/numpy/numpy/blob/df256d0d2f3bc6833699529824781c58f9c6e697/numpy/core/src/multiarray/flagsobject.c#L95C1-L99C45:
-    //
-    // The traditional rule is that for an array to be flagged as C contiguous,
-    // the following must hold:
-    //
-    // strides[-1] == itemsize
-    // strides[i] == shape[i+1] * strides[i + 1]
-    // [...]
-    // According to these rules, a 0- or 1-dimensional array is either both
-    // C- and F-contiguous, or neither; and an array with 2+ dimensions
-    // can be C- or F- contiguous, or neither, but not both. Though there
-    // there are exceptions for arrays with zero or one item, in the first
-    // case the check is relaxed up to and including the first dimension
-    // with shape[i] == 0. In the second case `strides == itemsize` will
-    // can be true for all dimensions and both flags are set.
-
-    if (ndarray->ndims == 0) {
-        return true;
-    }
-
-    if (ndarray->strides[ndarray->ndims - 1] != ndarray->itemsize) {
-        return false;
-    }
-
-    for (SizeT i = 1; i < ndarray->ndims; i++) {
-        SizeT axis_i = ndarray->ndims - i - 1;
-        if (ndarray->strides[axis_i] != ndarray->shape[axis_i + 1] * ndarray->strides[axis_i + 1]) {
-            return false;
-        }
-    }
-
-    return true;
-}
-
-/**
- * @brief Return the pointer to the element indexed by `indices` along the ndarray's axes.
- *
- * This function does no bound check.
- */
-template<typename SizeT>
-void* get_pelement_by_indices(const NDArray<SizeT>* ndarray, const SizeT* indices) {
-    void* element = ndarray->data;
-    for (SizeT dim_i = 0; dim_i < ndarray->ndims; dim_i++)
-        element = static_cast<uint8_t*>(element) + indices[dim_i] * ndarray->strides[dim_i];
-    return element;
-}
-
-/**
- * @brief Return the pointer to the nth (0-based) element of `ndarray` in flattened view.
- *
- * This function does no bound check.
- */
-template<typename SizeT>
-void* get_nth_pelement(const NDArray<SizeT>* ndarray, SizeT nth) {
-    void* element = ndarray->data;
-    for (SizeT i = 0; i < ndarray->ndims; i++) {
-        SizeT axis = ndarray->ndims - i - 1;
-        SizeT dim = ndarray->shape[axis];
-        element = static_cast<uint8_t*>(element) + ndarray->strides[axis] * (nth % dim);
-        nth /= dim;
-    }
-    return element;
-}
-
-/**
- * @brief Update the strides of an ndarray given an ndarray `shape` to be contiguous.
- *
- * You might want to read https://ajcr.net/stride-guide-part-1/.
- */
-template<typename SizeT>
-void set_strides_by_shape(NDArray<SizeT>* ndarray) {
-    SizeT stride_product = 1;
-    for (SizeT i = 0; i < ndarray->ndims; i++) {
-        SizeT axis = ndarray->ndims - i - 1;
-        ndarray->strides[axis] = stride_product * ndarray->itemsize;
-        stride_product *= ndarray->shape[axis];
-    }
-}
-
-/**
- * @brief Set an element in `ndarray`.
- *
- * @param pelement Pointer to the element in `ndarray` to be set.
- * @param pvalue Pointer to the value `pelement` will be set to.
- */
-template<typename SizeT>
-void set_pelement_value(NDArray<SizeT>* ndarray, void* pelement, const void* pvalue) {
-    __builtin_memcpy(pelement, pvalue, ndarray->itemsize);
-}
-
-/**
- * @brief Copy data from one ndarray to another of the exact same size and itemsize.
- *
- * Both ndarrays will be viewed in their flatten views when copying the elements.
- */
-template<typename SizeT>
-void copy_data(const NDArray<SizeT>* src_ndarray, NDArray<SizeT>* dst_ndarray) {
-    // TODO: Make this faster with memcpy when we see a contiguous segment.
-    // TODO: Handle overlapping.
-
-    debug_assert_eq(SizeT, src_ndarray->itemsize, dst_ndarray->itemsize);
-
-    for (SizeT i = 0; i < size(src_ndarray); i++) {
-        auto src_element = ndarray::basic::get_nth_pelement(src_ndarray, i);
-        auto dst_element = ndarray::basic::get_nth_pelement(dst_ndarray, i);
-        ndarray::basic::set_pelement_value(dst_ndarray, dst_element, src_element);
-    }
-}
-}  // namespace ndarray::basic
-}  // namespace
-
-extern "C" {
-using namespace ndarray::basic;
-
-void __nac3_ndarray_util_assert_shape_no_negative(int32_t ndims, int32_t* shape) {
-    assert_shape_no_negative(ndims, shape);
-}
-
-void __nac3_ndarray_util_assert_shape_no_negative64(int64_t ndims, int64_t* shape) {
-    assert_shape_no_negative(ndims, shape);
-}
-
-void __nac3_ndarray_util_assert_output_shape_same(int32_t ndarray_ndims,
-                                                  const int32_t* ndarray_shape,
-                                                  int32_t output_ndims,
-                                                  const int32_t* output_shape) {
-    assert_output_shape_same(ndarray_ndims, ndarray_shape, output_ndims, output_shape);
-}
-
-void __nac3_ndarray_util_assert_output_shape_same64(int64_t ndarray_ndims,
-                                                    const int64_t* ndarray_shape,
-                                                    int64_t output_ndims,
-                                                    const int64_t* output_shape) {
-    assert_output_shape_same(ndarray_ndims, ndarray_shape, output_ndims, output_shape);
-}
-
-uint32_t __nac3_ndarray_size(NDArray<int32_t>* ndarray) {
-    return size(ndarray);
-}
-
-uint64_t __nac3_ndarray_size64(NDArray<int64_t>* ndarray) {
-    return size(ndarray);
-}
-
-uint32_t __nac3_ndarray_nbytes(NDArray<int32_t>* ndarray) {
-    return nbytes(ndarray);
-}
-
-uint64_t __nac3_ndarray_nbytes64(NDArray<int64_t>* ndarray) {
-    return nbytes(ndarray);
-}
-
-int32_t __nac3_ndarray_len(NDArray<int32_t>* ndarray) {
-    return len(ndarray);
-}
-
-int64_t __nac3_ndarray_len64(NDArray<int64_t>* ndarray) {
-    return len(ndarray);
-}
-
-bool __nac3_ndarray_is_c_contiguous(NDArray<int32_t>* ndarray) {
-    return is_c_contiguous(ndarray);
-}
-
-bool __nac3_ndarray_is_c_contiguous64(NDArray<int64_t>* ndarray) {
-    return is_c_contiguous(ndarray);
-}
-
-void* __nac3_ndarray_get_nth_pelement(const NDArray<int32_t>* ndarray, int32_t nth) {
-    return get_nth_pelement(ndarray, nth);
-}
-
-void* __nac3_ndarray_get_nth_pelement64(const NDArray<int64_t>* ndarray, int64_t nth) {
-    return get_nth_pelement(ndarray, nth);
-}
-
-void* __nac3_ndarray_get_pelement_by_indices(const NDArray<int32_t>* ndarray, int32_t* indices) {
-    return get_pelement_by_indices(ndarray, indices);
-}
-
-void* __nac3_ndarray_get_pelement_by_indices64(const NDArray<int64_t>* ndarray, int64_t* indices) {
-    return get_pelement_by_indices(ndarray, indices);
-}
-
-void __nac3_ndarray_set_strides_by_shape(NDArray<int32_t>* ndarray) {
-    set_strides_by_shape(ndarray);
-}
-
-void __nac3_ndarray_set_strides_by_shape64(NDArray<int64_t>* ndarray) {
-    set_strides_by_shape(ndarray);
-}
-
-void __nac3_ndarray_copy_data(NDArray<int32_t>* src_ndarray, NDArray<int32_t>* dst_ndarray) {
-    copy_data(src_ndarray, dst_ndarray);
-}
-
-void __nac3_ndarray_copy_data64(NDArray<int64_t>* src_ndarray, NDArray<int64_t>* dst_ndarray) {
-    copy_data(src_ndarray, dst_ndarray);
-}
-}

nac3core/irrt/irrt/ndarray/broadcast.hpp

@@ -1,165 +0,0 @@
-#pragma once
-
-#include "irrt/int_types.hpp"
-#include "irrt/ndarray/def.hpp"
-#include "irrt/slice.hpp"
-
-namespace {
-template<typename SizeT>
-struct ShapeEntry {
-    SizeT ndims;
-    SizeT* shape;
-};
-}  // namespace
-
-namespace {
-namespace ndarray::broadcast {
-/**
- * @brief Return true if `src_shape` can broadcast to `dst_shape`.
- *
- * See https://numpy.org/doc/stable/user/basics.broadcasting.html
- */
-template<typename SizeT>
-bool can_broadcast_shape_to(SizeT target_ndims, const SizeT* target_shape, SizeT src_ndims, const SizeT* src_shape) {
-    if (src_ndims > target_ndims) {
-        return false;
-    }
-
-    for (SizeT i = 0; i < src_ndims; i++) {
-        SizeT target_dim = target_shape[target_ndims - i - 1];
-        SizeT src_dim = src_shape[src_ndims - i - 1];
-        if (!(src_dim == 1 || target_dim == src_dim)) {
-            return false;
-        }
-    }
-    return true;
-}
-
-/**
- * @brief Performs `np.broadcast_shapes(<shapes>)`
- *
- * @param num_shapes Number of entries in `shapes`
- * @param shapes The list of shape to do `np.broadcast_shapes` on.
- * @param dst_ndims The length of `dst_shape`.
- *   `dst_ndims` must be `max([shape.ndims for shape in shapes])`, but the caller has to calculate it/provide it.
- *   for this function since they should already know in order to allocate `dst_shape` in the first place.
- * @param dst_shape The resulting shape. Must be pre-allocated by the caller. This function calculate the result
- *   of `np.broadcast_shapes` and write it here.
- */
-template<typename SizeT>
-void broadcast_shapes(SizeT num_shapes, const ShapeEntry<SizeT>* shapes, SizeT dst_ndims, SizeT* dst_shape) {
-    for (SizeT dst_axis = 0; dst_axis < dst_ndims; dst_axis++) {
-        dst_shape[dst_axis] = 1;
-    }
-
-#ifdef IRRT_DEBUG_ASSERT
-    SizeT max_ndims_found = 0;
-#endif
-
-    for (SizeT i = 0; i < num_shapes; i++) {
-        ShapeEntry<SizeT> entry = shapes[i];
-
-        // Check pre-condition: `dst_ndims` must be `max([shape.ndims for shape in shapes])`
-        debug_assert(SizeT, entry.ndims <= dst_ndims);
-
-#ifdef IRRT_DEBUG_ASSERT
-        max_ndims_found = max(max_ndims_found, entry.ndims);
-#endif
-
-        for (SizeT j = 0; j < entry.ndims; j++) {
-            SizeT entry_axis = entry.ndims - j - 1;
-            SizeT dst_axis = dst_ndims - j - 1;
-
-            SizeT entry_dim = entry.shape[entry_axis];
-            SizeT dst_dim = dst_shape[dst_axis];
-
-            if (dst_dim == 1) {
-                dst_shape[dst_axis] = entry_dim;
-            } else if (entry_dim == 1 || entry_dim == dst_dim) {
-                // Do nothing
-            } else {
-                raise_exception(SizeT, EXN_VALUE_ERROR,
-                                "shape mismatch: objects cannot be broadcast "
-                                "to a single shape.",
-                                NO_PARAM, NO_PARAM, NO_PARAM);
-            }
-        }
-    }
-
-#ifdef IRRT_DEBUG_ASSERT
-    // Check pre-condition: `dst_ndims` must be `max([shape.ndims for shape in shapes])`
-    debug_assert_eq(SizeT, max_ndims_found, dst_ndims);
-#endif
-}
-
-/**
- * @brief Perform `np.broadcast_to(<ndarray>, <target_shape>)` and appropriate assertions.
- *
- * This function attempts to broadcast `src_ndarray` to a new shape defined by `dst_ndarray.shape`,
- * and return the result by modifying `dst_ndarray`.
- *
- * # Notes on `dst_ndarray`
- * The caller is responsible for allocating space for the resulting ndarray.
- * Here is what this function expects from `dst_ndarray` when called:
- *   - `dst_ndarray->data` does not have to be initialized.
- *   - `dst_ndarray->itemsize` does not have to be initialized.
- *   - `dst_ndarray->ndims` must be initialized, determining the length of `dst_ndarray->shape`
- *   - `dst_ndarray->shape` must be allocated, and must contain the desired target broadcast shape.
- *   - `dst_ndarray->strides` must be allocated, through it can contain uninitialized values.
- * When this function call ends:
- *   - `dst_ndarray->data` is set to `src_ndarray->data` (`dst_ndarray` is just a view to `src_ndarray`)
- *   - `dst_ndarray->itemsize` is set to `src_ndarray->itemsize`
- *   - `dst_ndarray->ndims` is unchanged.
- *   - `dst_ndarray->shape` is unchanged.
- *   - `dst_ndarray->strides` is updated accordingly by how ndarray broadcast_to works.
- */
-template<typename SizeT>
-void broadcast_to(const NDArray<SizeT>* src_ndarray, NDArray<SizeT>* dst_ndarray) {
-    if (!ndarray::broadcast::can_broadcast_shape_to(dst_ndarray->ndims, dst_ndarray->shape, src_ndarray->ndims,
-                                                    src_ndarray->shape)) {
-        raise_exception(SizeT, EXN_VALUE_ERROR, "operands could not be broadcast together", NO_PARAM, NO_PARAM,
-                        NO_PARAM);
-    }
-
-    dst_ndarray->data = src_ndarray->data;
-    dst_ndarray->itemsize = src_ndarray->itemsize;
-
-    for (SizeT i = 0; i < dst_ndarray->ndims; i++) {
-        SizeT src_axis = src_ndarray->ndims - i - 1;
-        SizeT dst_axis = dst_ndarray->ndims - i - 1;
-        if (src_axis < 0 || (src_ndarray->shape[src_axis] == 1 && dst_ndarray->shape[dst_axis] != 1)) {
-            // Freeze the steps in-place
-            dst_ndarray->strides[dst_axis] = 0;
-        } else {
-            dst_ndarray->strides[dst_axis] = src_ndarray->strides[src_axis];
-        }
-    }
-}
-}  // namespace ndarray::broadcast
-}  // namespace
-
-extern "C" {
-using namespace ndarray::broadcast;
-
-void __nac3_ndarray_broadcast_to(NDArray<int32_t>* src_ndarray, NDArray<int32_t>* dst_ndarray) {
-    broadcast_to(src_ndarray, dst_ndarray);
-}
-
-void __nac3_ndarray_broadcast_to64(NDArray<int64_t>* src_ndarray, NDArray<int64_t>* dst_ndarray) {
-    broadcast_to(src_ndarray, dst_ndarray);
-}
-
-void __nac3_ndarray_broadcast_shapes(int32_t num_shapes,
-                                     const ShapeEntry<int32_t>* shapes,
-                                     int32_t dst_ndims,
-                                     int32_t* dst_shape) {
-    broadcast_shapes(num_shapes, shapes, dst_ndims, dst_shape);
-}
-
-void __nac3_ndarray_broadcast_shapes64(int64_t num_shapes,
-                                       const ShapeEntry<int64_t>* shapes,
-                                       int64_t dst_ndims,
-                                       int64_t* dst_shape) {
-    broadcast_shapes(num_shapes, shapes, dst_ndims, dst_shape);
-}
-}

nac3core/irrt/irrt/ndarray/def.hpp

@@ -1,51 +0,0 @@
-#pragma once
-
-#include "irrt/int_types.hpp"
-
-namespace {
-/**
- * @brief The NDArray object
- *
- * Official numpy implementation:
- * https://github.com/numpy/numpy/blob/735a477f0bc2b5b84d0e72d92f224bde78d4e069/doc/source/reference/c-api/types-and-structures.rst#pyarrayinterface
- *
- * Note that this implementation is based on `PyArrayInterface` rather of `PyArrayObject`. The
- * difference between `PyArrayInterface` and `PyArrayObject` (relevant to our implementation) is
- * that `PyArrayInterface` *has* `itemsize` and uses `void*` for its `data`, whereas `PyArrayObject`
- * does not require `itemsize` (probably using `strides[-1]` instead) and uses `char*` for its
- * `data`. There are also minor differences in the struct layout.
- */
-template<typename SizeT>
-struct NDArray {
-    /**
-     * @brief The number of bytes of a single element in `data`.
-     */
-    SizeT itemsize;
-
-    /**
-     * @brief The number of dimensions of this shape.
-     */
-    SizeT ndims;
-
-    /**
-     * @brief The NDArray shape, with length equal to `ndims`.
-     *
-     * Note that it may contain 0.
-     */
-    SizeT* shape;
-
-    /**
-     * @brief Array strides, with length equal to `ndims`
-     *
-     * The stride values are in units of bytes, not number of elements.
-     *
-     * Note that `strides` can have negative values or contain 0.
-     */
-    SizeT* strides;
-
-    /**
-     * @brief The underlying data this `ndarray` is pointing to.
-     */
-    void* data;
-};
-}  // namespace

nac3core/irrt/irrt/ndarray/indexing.hpp

@@ -1,219 +0,0 @@
-#pragma once
-
-#include "irrt/exception.hpp"
-#include "irrt/int_types.hpp"
-#include "irrt/ndarray/basic.hpp"
-#include "irrt/ndarray/def.hpp"
-#include "irrt/range.hpp"
-#include "irrt/slice.hpp"
-
-namespace {
-typedef uint8_t NDIndexType;
-
-/**
- * @brief A single element index
- *
- * `data` points to a `int32_t`.
- */
-const NDIndexType ND_INDEX_TYPE_SINGLE_ELEMENT = 0;
-
-/**
- * @brief A slice index
- *
- * `data` points to a `Slice<int32_t>`.
- */
-const NDIndexType ND_INDEX_TYPE_SLICE = 1;
-
-/**
- * @brief `np.newaxis` / `None`
- *
- * `data` is unused.
- */
-const NDIndexType ND_INDEX_TYPE_NEWAXIS = 2;
-
-/**
- * @brief `Ellipsis` / `...`
- *
- * `data` is unused.
- */
-const NDIndexType ND_INDEX_TYPE_ELLIPSIS = 3;
-
-/**
- * @brief An index used in ndarray indexing
- *
- * That is:
- * ```
- * my_ndarray[::-1, 3, ..., np.newaxis]
- *            ^^^^  ^  ^^^  ^^^^^^^^^^ each of these is represented by an NDIndex.
- * ```
- */
-struct NDIndex {
-    /**
-     * @brief Enum tag to specify the type of index.
-     *
-     * Please see the comment of each enum constant.
-     */
-    NDIndexType type;
-
-    /**
-     * @brief The accompanying data associated with `type`.
-     *
-     * Please see the comment of each enum constant.
-     */
-    uint8_t* data;
-};
-}  // namespace
-
-namespace {
-namespace ndarray::indexing {
-/**
- * @brief Perform ndarray "basic indexing" (https://numpy.org/doc/stable/user/basics.indexing.html#basic-indexing)
- *
- * This function is very similar to performing `dst_ndarray = src_ndarray[indices]` in Python.
- *
- * This function also does proper assertions on `indices` to check for out of bounds access and more.
- *
- * # Notes on `dst_ndarray`
- * The caller is responsible for allocating space for the resulting ndarray.
- * Here is what this function expects from `dst_ndarray` when called:
- *   - `dst_ndarray->data` does not have to be initialized.
- *   - `dst_ndarray->itemsize` does not have to be initialized.
- *   - `dst_ndarray->ndims` must be initialized, and it must be equal to the expected `ndims` of the `dst_ndarray` after
- *       indexing `src_ndarray` with `indices`.
- *   - `dst_ndarray->shape` must be allocated, through it can contain uninitialized values.
- *   - `dst_ndarray->strides` must be allocated, through it can contain uninitialized values.
- * When this function call ends:
- *   - `dst_ndarray->data` is set to `src_ndarray->data`.
- *   - `dst_ndarray->itemsize` is set to `src_ndarray->itemsize`.
- *   - `dst_ndarray->ndims` is unchanged.
- *   - `dst_ndarray->shape` is updated according to how `src_ndarray` is indexed.
- *   - `dst_ndarray->strides` is updated accordingly by how ndarray indexing works.
- *
- * @param indices indices to index `src_ndarray`, ordered in the same way you would write them in Python.
- * @param src_ndarray The NDArray to be indexed.
- * @param dst_ndarray The resulting NDArray after indexing. Further details in the comments above,
- */
-template<typename SizeT>
-void index(SizeT num_indices, const NDIndex* indices, const NDArray<SizeT>* src_ndarray, NDArray<SizeT>* dst_ndarray) {
-    // Validate `indices`.
-
-    // Expected value of `dst_ndarray->ndims`.
-    SizeT expected_dst_ndims = src_ndarray->ndims;
-    // To check for "too many indices for array: array is ?-dimensional, but ? were indexed"
-    SizeT num_indexed = 0;
-    // There may be ellipsis `...` in `indices`. There can only be 0 or 1 ellipsis.
-    SizeT num_ellipsis = 0;
-
-    for (SizeT i = 0; i < num_indices; i++) {
-        if (indices[i].type == ND_INDEX_TYPE_SINGLE_ELEMENT) {
-            expected_dst_ndims--;
-            num_indexed++;
-        } else if (indices[i].type == ND_INDEX_TYPE_SLICE) {
-            num_indexed++;
-        } else if (indices[i].type == ND_INDEX_TYPE_NEWAXIS) {
-            expected_dst_ndims++;
-        } else if (indices[i].type == ND_INDEX_TYPE_ELLIPSIS) {
-            num_ellipsis++;
-            if (num_ellipsis > 1) {
-                raise_exception(SizeT, EXN_INDEX_ERROR, "an index can only have a single ellipsis ('...')", NO_PARAM,
-                                NO_PARAM, NO_PARAM);
-            }
-        } else {
-            __builtin_unreachable();
-        }
-    }
-
-    debug_assert_eq(SizeT, expected_dst_ndims, dst_ndarray->ndims);
-
-    if (src_ndarray->ndims - num_indexed < 0) {
-        raise_exception(SizeT, EXN_INDEX_ERROR,
-                        "too many indices for array: array is {0}-dimensional, "
-                        "but {1} were indexed",
-                        src_ndarray->ndims, num_indices, NO_PARAM);
-    }
-
-    dst_ndarray->data = src_ndarray->data;
-    dst_ndarray->itemsize = src_ndarray->itemsize;
-
-    // Reference code:
-    // https://github.com/wadetb/tinynumpy/blob/0d23d22e07062ffab2afa287374c7b366eebdda1/tinynumpy/tinynumpy.py#L652
-    SizeT src_axis = 0;
-    SizeT dst_axis = 0;
-
-    for (int32_t i = 0; i < num_indices; i++) {
-        const NDIndex* index = &indices[i];
-        if (index->type == ND_INDEX_TYPE_SINGLE_ELEMENT) {
-            SizeT input = (SizeT) * ((int32_t*)index->data);
-
-            SizeT k = slice::resolve_index_in_length(src_ndarray->shape[src_axis], input);
-            if (k == -1) {
-                raise_exception(SizeT, EXN_INDEX_ERROR,
-                                "index {0} is out of bounds for axis {1} "
-                                "with size {2}",
-                                input, src_axis, src_ndarray->shape[src_axis]);
-            }
-
-            dst_ndarray->data = static_cast<uint8_t*>(dst_ndarray->data) + k * src_ndarray->strides[src_axis];
-
-            src_axis++;
-        } else if (index->type == ND_INDEX_TYPE_SLICE) {
-            Slice<int32_t>* slice = (Slice<int32_t>*)index->data;
-
-            Range<int32_t> range = slice->indices_checked<SizeT>(src_ndarray->shape[src_axis]);
-
-            dst_ndarray->data =
-                static_cast<uint8_t*>(dst_ndarray->data) + (SizeT)range.start * src_ndarray->strides[src_axis];
-            dst_ndarray->strides[dst_axis] = ((SizeT)range.step) * src_ndarray->strides[src_axis];
-            dst_ndarray->shape[dst_axis] = (SizeT)range.len<SizeT>();
-
-            dst_axis++;
-            src_axis++;
-        } else if (index->type == ND_INDEX_TYPE_NEWAXIS) {
-            dst_ndarray->strides[dst_axis] = 0;
-            dst_ndarray->shape[dst_axis] = 1;
-
-            dst_axis++;
-        } else if (index->type == ND_INDEX_TYPE_ELLIPSIS) {
-            // The number of ':' entries this '...' implies.
-            SizeT ellipsis_size = src_ndarray->ndims - num_indexed;
-
-            for (SizeT j = 0; j < ellipsis_size; j++) {
-                dst_ndarray->strides[dst_axis] = src_ndarray->strides[src_axis];
-                dst_ndarray->shape[dst_axis] = src_ndarray->shape[src_axis];
-
-                dst_axis++;
-                src_axis++;
-            }
-        } else {
-            __builtin_unreachable();
-        }
-    }
-
-    for (; dst_axis < dst_ndarray->ndims; dst_axis++, src_axis++) {
-        dst_ndarray->shape[dst_axis] = src_ndarray->shape[src_axis];
-        dst_ndarray->strides[dst_axis] = src_ndarray->strides[src_axis];
-    }
-
-    debug_assert_eq(SizeT, src_ndarray->ndims, src_axis);
-    debug_assert_eq(SizeT, dst_ndarray->ndims, dst_axis);
-}
-}  // namespace ndarray::indexing
-}  // namespace
-
-extern "C" {
-using namespace ndarray::indexing;
-
-void __nac3_ndarray_index(int32_t num_indices,
-                          NDIndex* indices,
-                          NDArray<int32_t>* src_ndarray,
-                          NDArray<int32_t>* dst_ndarray) {
-    index(num_indices, indices, src_ndarray, dst_ndarray);
-}
-
-void __nac3_ndarray_index64(int64_t num_indices,
-                            NDIndex* indices,
-                            NDArray<int64_t>* src_ndarray,
-                            NDArray<int64_t>* dst_ndarray) {
-    index(num_indices, indices, src_ndarray, dst_ndarray);
-}
-}

nac3core/irrt/irrt/ndarray/iter.hpp

@@ -1,146 +0,0 @@
-#pragma once
-
-#include "irrt/int_types.hpp"
-#include "irrt/ndarray/def.hpp"
-
-namespace {
-/**
- * @brief Helper struct to enumerate through an ndarray *efficiently*.
- *
- * Example usage (in pseudo-code):
- * ```
- * // Suppose my_ndarray has been initialized, with shape [2, 3] and dtype `double`
- * NDIter nditer;
- * nditer.initialize(my_ndarray);
- * while (nditer.has_element()) {
- *    // This body is run 6 (= my_ndarray.size) times.
- *
- *    // [0, 0] -> [0, 1] -> [0, 2] -> [1, 0] -> [1, 1] -> [1, 2] -> end
- *    print(nditer.indices);
- *
- *    // 0 -> 1 -> 2 -> 3 -> 4 -> 5
- *    print(nditer.nth);
- *
- *    // <1st element> -> <2nd element> -> ... -> <6th element> -> end
- *    print(*((double *) nditer.element))
- *
- *    nditer.next(); // Go to next element.
- * }
- * ```
- *
- * Interesting cases:
- * - If `my_ndarray.ndims` == 0, there is one iteration.
- * - If `my_ndarray.shape` contains zeroes, there are no iterations.
- */
-template<typename SizeT>
-struct NDIter {
-    // Information about the ndarray being iterated over.
-    SizeT ndims;
-    SizeT* shape;
-    SizeT* strides;
-
-    /**
-     * @brief The current indices.
-     *
-     * Must be allocated by the caller.
-     */
-    SizeT* indices;
-
-    /**
-     * @brief The nth (0-based) index of the current indices.
-     *
-     * Initially this is 0.
-     */
-    SizeT nth;
-
-    /**
-     * @brief Pointer to the current element.
-     *
-     * Initially this points to first element of the ndarray.
-     */
-    void* element;
-
-    /**
-     * @brief Cache for the product of shape.
-     *
-     * Could be 0 if `shape` has 0s in it.
-     */
-    SizeT size;
-
-    void initialize(SizeT ndims, SizeT* shape, SizeT* strides, void* element, SizeT* indices) {
-        this->ndims = ndims;
-        this->shape = shape;
-        this->strides = strides;
-
-        this->indices = indices;
-        this->element = element;
-
-        // Compute size
-        this->size = 1;
-        for (SizeT i = 0; i < ndims; i++) {
-            this->size *= shape[i];
-        }
-
-        // `indices` starts on all 0s.
-        for (SizeT axis = 0; axis < ndims; axis++)
-            indices[axis] = 0;
-        nth = 0;
-    }
-
-    void initialize_by_ndarray(NDArray<SizeT>* ndarray, SizeT* indices) {
-        // NOTE: ndarray->data is pointing to the first element, and `NDIter`'s `element` should also point to the first
-        // element as well.
-        this->initialize(ndarray->ndims, ndarray->shape, ndarray->strides, ndarray->data, indices);
-    }
-
-    // Is the current iteration valid?
-    // If true, then `element`, `indices` and `nth` contain details about the current element.
-    bool has_element() { return nth < size; }
-
-    // Go to the next element.
-    void next() {
-        for (SizeT i = 0; i < ndims; i++) {
-            SizeT axis = ndims - i - 1;
-            indices[axis]++;
-            if (indices[axis] >= shape[axis]) {
-                indices[axis] = 0;
-
-                // TODO: There is something called backstrides to speedup iteration.
-                // See https://ajcr.net/stride-guide-part-1/, and
-                // https://docs.scipy.org/doc/numpy-1.13.0/reference/c-api.types-and-structures.html#c.PyArrayIterObject.PyArrayIterObject.backstrides.
-                element = static_cast<void*>(reinterpret_cast<uint8_t*>(element) - strides[axis] * (shape[axis] - 1));
-            } else {
-                element = static_cast<void*>(reinterpret_cast<uint8_t*>(element) + strides[axis]);
-                break;
-            }
-        }
-        nth++;
-    }
-};
-}  // namespace
-
-extern "C" {
-void __nac3_nditer_initialize(NDIter<int32_t>* iter, NDArray<int32_t>* ndarray, int32_t* indices) {
-    iter->initialize_by_ndarray(ndarray, indices);
-}
-
-void __nac3_nditer_initialize64(NDIter<int64_t>* iter, NDArray<int64_t>* ndarray, int64_t* indices) {
-    iter->initialize_by_ndarray(ndarray, indices);
-}
-
-bool __nac3_nditer_has_element(NDIter<int32_t>* iter) {
-    return iter->has_element();
-}
-
-bool __nac3_nditer_has_element64(NDIter<int64_t>* iter) {
-    return iter->has_element();
-}
-
-void __nac3_nditer_next(NDIter<int32_t>* iter) {
-    iter->next();
-}
-
-void __nac3_nditer_next64(NDIter<int64_t>* iter) {
-    iter->next();
-}
-}

nac3core/irrt/irrt/ndarray/matmul.hpp

@@ -1,98 +0,0 @@
-#pragma once
-
-#include "irrt/debug.hpp"
-#include "irrt/exception.hpp"
-#include "irrt/int_types.hpp"
-#include "irrt/ndarray/basic.hpp"
-#include "irrt/ndarray/broadcast.hpp"
-#include "irrt/ndarray/iter.hpp"
-
-// NOTE: Everything would be much easier and elegant if einsum is implemented.
-
-namespace {
-namespace ndarray::matmul {
-
-/**
- * @brief Perform the broadcast in `np.einsum("...ij,...jk->...ik", a, b)`.
- *
- * Example:
- *   Suppose      `a_shape ==      [1, 97, 4, 2]`
- *       and      `b_shape == [99, 98,  1, 2, 5]`,
- *
- *   ...then  `new_a_shape == [99, 98, 97, 4, 2]`,
- *            `new_b_shape == [99, 98, 97, 2, 5]`,
- *      and     `dst_shape == [99, 98, 97, 4, 5]`.
- *                             ^^^^^^^^^^  ^^^^
- *                          (broadcasted)  (4x2 @ 2x5 => 4x5)
- *
- * @param a_ndims Length of `a_shape`.
- * @param a_shape Shape of `a`.
- * @param b_ndims Length of `b_shape`.
- * @param b_shape Shape of `b`.
- * @param final_ndims Should be equal to `max(a_ndims, b_ndims)`. This is the length of `new_a_shape`,
- * `new_b_shape`, and `dst_shape` - the number of dimensions after broadcasting.
- */
-template<typename SizeT>
-void calculate_shapes(SizeT a_ndims,
-                      SizeT* a_shape,
-                      SizeT b_ndims,
-                      SizeT* b_shape,
-                      SizeT final_ndims,
-                      SizeT* new_a_shape,
-                      SizeT* new_b_shape,
-                      SizeT* dst_shape) {
-    debug_assert(SizeT, a_ndims >= 2);
-    debug_assert(SizeT, b_ndims >= 2);
-    debug_assert_eq(SizeT, max(a_ndims, b_ndims), final_ndims);
-
-    // Check that a and b are compatible for matmul
-    if (a_shape[a_ndims - 1] != b_shape[b_ndims - 2]) {
-        // This is a custom error message. Different from NumPy.
-        raise_exception(SizeT, EXN_VALUE_ERROR, "Cannot multiply LHS (shape ?x{0}) with RHS (shape {1}x?})",
-                        a_shape[a_ndims - 1], b_shape[b_ndims - 2], NO_PARAM);
-    }
-
-    const SizeT num_entries = 2;
-    ShapeEntry<SizeT> entries[num_entries] = {{.ndims = a_ndims - 2, .shape = a_shape},
-                                              {.ndims = b_ndims - 2, .shape = b_shape}};
-
-    // TODO: Optimize this
-    ndarray::broadcast::broadcast_shapes<SizeT>(num_entries, entries, final_ndims - 2, new_a_shape);
-    ndarray::broadcast::broadcast_shapes<SizeT>(num_entries, entries, final_ndims - 2, new_b_shape);
-    ndarray::broadcast::broadcast_shapes<SizeT>(num_entries, entries, final_ndims - 2, dst_shape);
-
-    new_a_shape[final_ndims - 2] = a_shape[a_ndims - 2];
-    new_a_shape[final_ndims - 1] = a_shape[a_ndims - 1];
-    new_b_shape[final_ndims - 2] = b_shape[b_ndims - 2];
-    new_b_shape[final_ndims - 1] = b_shape[b_ndims - 1];
-    dst_shape[final_ndims - 2] = a_shape[a_ndims - 2];
-    dst_shape[final_ndims - 1] = b_shape[b_ndims - 1];
-}
-}  // namespace ndarray::matmul
-}  // namespace
-
-extern "C" {
-using namespace ndarray::matmul;
-
-void __nac3_ndarray_matmul_calculate_shapes(int32_t a_ndims,
-                                            int32_t* a_shape,
-                                            int32_t b_ndims,
-                                            int32_t* b_shape,
-                                            int32_t final_ndims,
-                                            int32_t* new_a_shape,
-                                            int32_t* new_b_shape,
-                                            int32_t* dst_shape) {
-    calculate_shapes(a_ndims, a_shape, b_ndims, b_shape, final_ndims, new_a_shape, new_b_shape, dst_shape);
-}
-
-void __nac3_ndarray_matmul_calculate_shapes64(int64_t a_ndims,
-                                              int64_t* a_shape,
-                                              int64_t b_ndims,
-                                              int64_t* b_shape,
-                                              int64_t final_ndims,
-                                              int64_t* new_a_shape,
-                                              int64_t* new_b_shape,
-                                              int64_t* dst_shape) {
-    calculate_shapes(a_ndims, a_shape, b_ndims, b_shape, final_ndims, new_a_shape, new_b_shape, dst_shape);
-}
-}

nac3core/irrt/irrt/ndarray/reshape.hpp

@@ -1,97 +0,0 @@
-#pragma once
-
-#include "irrt/exception.hpp"
-#include "irrt/int_types.hpp"
-#include "irrt/ndarray/def.hpp"
-
-namespace {
-namespace ndarray::reshape {
-/**
- * @brief Perform assertions on and resolve unknown dimensions in `new_shape` in `np.reshape(<ndarray>, new_shape)`
- *
- * If `new_shape` indeed contains unknown dimensions (specified with `-1`, just like numpy), `new_shape` will be
- * modified to contain the resolved dimension.
- *
- * To perform assertions on and resolve unknown dimensions in `new_shape`, we don't need the actual
- * `<ndarray>` object itself, but only the `.size` of the `<ndarray>`.
- *
- * @param size The `.size` of `<ndarray>`
- * @param new_ndims Number of elements in `new_shape`
- * @param new_shape Target shape to reshape to
- */
-template<typename SizeT>
-void resolve_and_check_new_shape(SizeT size, SizeT new_ndims, SizeT* new_shape) {
-    // Is there a -1 in `new_shape`?
-    bool neg1_exists = false;
-    // Location of -1, only initialized if `neg1_exists` is true
-    SizeT neg1_axis_i;
-    // The computed ndarray size of `new_shape`
-    SizeT new_size = 1;
-
-    for (SizeT axis_i = 0; axis_i < new_ndims; axis_i++) {
-        SizeT dim = new_shape[axis_i];
-        if (dim < 0) {
-            if (dim == -1) {
-                if (neg1_exists) {
-                    // Multiple `-1` found. Throw an error.
-                    raise_exception(SizeT, EXN_VALUE_ERROR, "can only specify one unknown dimension", NO_PARAM,
-                                    NO_PARAM, NO_PARAM);
-                } else {
-                    neg1_exists = true;
-                    neg1_axis_i = axis_i;
-                }
-            } else {
-                // TODO: What? In `np.reshape` any negative dimensions is
-                // treated like its `-1`.
-                //
-                // Try running `np.zeros((3, 4)).reshape((-999, 2))`
-                //
-                // It is not documented by numpy.
-                // Throw an error for now...
-
-                raise_exception(SizeT, EXN_VALUE_ERROR, "Found non -1 negative dimension {0} on axis {1}", dim, axis_i,
-                                NO_PARAM);
-            }
-        } else {
-            new_size *= dim;
-        }
-    }
-
-    bool can_reshape;
-    if (neg1_exists) {
-        // Let `x` be the unknown dimension
-        // Solve `x * <new_size> = <size>`
-        if (new_size == 0 && size == 0) {
-            // `x` has infinitely many solutions
-            can_reshape = false;
-        } else if (new_size == 0 && size != 0) {
-            // `x` has no solutions
-            can_reshape = false;
-        } else if (size % new_size != 0) {
-            // `x` has no integer solutions
-            can_reshape = false;
-        } else {
-            can_reshape = true;
-            new_shape[neg1_axis_i] = size / new_size;  // Resolve dimension
-        }
-    } else {
-        can_reshape = (new_size == size);
-    }
-
-    if (!can_reshape) {
-        raise_exception(SizeT, EXN_VALUE_ERROR, "cannot reshape array of size {0} into given shape", size, NO_PARAM,
-                        NO_PARAM);
-    }
-}
-}  // namespace ndarray::reshape
-}  // namespace
-
-extern "C" {
-void __nac3_ndarray_reshape_resolve_and_check_new_shape(int32_t size, int32_t new_ndims, int32_t* new_shape) {
-    ndarray::reshape::resolve_and_check_new_shape(size, new_ndims, new_shape);
-}
-
-void __nac3_ndarray_reshape_resolve_and_check_new_shape64(int64_t size, int64_t new_ndims, int64_t* new_shape) {
-    ndarray::reshape::resolve_and_check_new_shape(size, new_ndims, new_shape);
-}
-}

nac3core/irrt/irrt/ndarray/transpose.hpp

@@ -1,143 +0,0 @@
-#pragma once
-
-#include "irrt/debug.hpp"
-#include "irrt/exception.hpp"
-#include "irrt/int_types.hpp"
-#include "irrt/ndarray/def.hpp"
-#include "irrt/slice.hpp"
-
-/*
- * Notes on `np.transpose(<array>, <axes>)`
- *
- * TODO: `axes`, if specified, can actually contain negative indices,
- * but it is not documented in numpy.
- *
- * Supporting it for now.
- */
-
-namespace {
-namespace ndarray::transpose {
-/**
- * @brief Do assertions on `<axes>` in `np.transpose(<array>, <axes>)`.
- *
- * Note that `np.transpose`'s `<axe>` argument is optional. If the argument
- * is specified but the user, use this function to do assertions on it.
- *
- * @param ndims The number of dimensions of `<array>`
- * @param num_axes Number of elements in `<axes>` as specified by the user.
- *   This should be equal to `ndims`. If not, a "ValueError: axes don't match array" is thrown.
- * @param axes The user specified `<axes>`.
- */
-template<typename SizeT>
-void assert_transpose_axes(SizeT ndims, SizeT num_axes, const SizeT* axes) {
-    if (ndims != num_axes) {
-        raise_exception(SizeT, EXN_VALUE_ERROR, "axes don't match array", NO_PARAM, NO_PARAM, NO_PARAM);
-    }
-
-    // TODO: Optimize this
-    bool* axe_specified = (bool*)__builtin_alloca(sizeof(bool) * ndims);
-    for (SizeT i = 0; i < ndims; i++)
-        axe_specified[i] = false;
-
-    for (SizeT i = 0; i < ndims; i++) {
-        SizeT axis = slice::resolve_index_in_length(ndims, axes[i]);
-        if (axis == -1) {
-            // TODO: numpy actually throws a `numpy.exceptions.AxisError`
-            raise_exception(SizeT, EXN_VALUE_ERROR, "axis {0} is out of bounds for array of dimension {1}", axis, ndims,
-                            NO_PARAM);
-        }
-
-        if (axe_specified[axis]) {
-            raise_exception(SizeT, EXN_VALUE_ERROR, "repeated axis in transpose", NO_PARAM, NO_PARAM, NO_PARAM);
-        }
-
-        axe_specified[axis] = true;
-    }
-}
-
-/**
- * @brief Create a transpose view of `src_ndarray` and perform proper assertions.
- *
- * This function is very similar to doing `dst_ndarray = np.transpose(src_ndarray, <axes>)`.
- * If `<axes>` is supposed to be `None`, caller can pass in a `nullptr` to `<axes>`.
- *
- * The transpose view created is returned by modifying `dst_ndarray`.
- *
- * The caller is responsible for setting up `dst_ndarray` before calling this function.
- * Here is what this function expects from `dst_ndarray` when called:
- *   - `dst_ndarray->data` does not have to be initialized.
- *   - `dst_ndarray->itemsize` does not have to be initialized.
- *   - `dst_ndarray->ndims` must be initialized, must be equal to `src_ndarray->ndims`.
- *   - `dst_ndarray->shape` must be allocated, through it can contain uninitialized values.
- *   - `dst_ndarray->strides` must be allocated, through it can contain uninitialized values.
- * When this function call ends:
- *   - `dst_ndarray->data` is set to `src_ndarray->data` (`dst_ndarray` is just a view to `src_ndarray`)
- *   - `dst_ndarray->itemsize` is set to `src_ndarray->itemsize`
- *   - `dst_ndarray->ndims` is unchanged
- *   - `dst_ndarray->shape` is updated according to how `np.transpose` works
- *   - `dst_ndarray->strides` is updated according to how `np.transpose` works
- *
- * @param src_ndarray The NDArray to build a transpose view on
- * @param dst_ndarray The resulting NDArray after transpose. Further details in the comments above,
- * @param num_axes Number of elements in axes. Unused if `axes` is nullptr.
- * @param axes Axes permutation. Set it to `nullptr` if `<axes>` is `None`.
- */
-template<typename SizeT>
-void transpose(const NDArray<SizeT>* src_ndarray, NDArray<SizeT>* dst_ndarray, SizeT num_axes, const SizeT* axes) {
-    debug_assert_eq(SizeT, src_ndarray->ndims, dst_ndarray->ndims);
-    const auto ndims = src_ndarray->ndims;
-
-    if (axes != nullptr)
-        assert_transpose_axes(ndims, num_axes, axes);
-
-    dst_ndarray->data = src_ndarray->data;
-    dst_ndarray->itemsize = src_ndarray->itemsize;
-
-    // Check out https://ajcr.net/stride-guide-part-2/ to see how `np.transpose` works behind the scenes.
-    if (axes == nullptr) {
-        // `np.transpose(<array>, axes=None)`
-
-        /*
-         * Minor note: `np.transpose(<array>, axes=None)` is equivalent to
-         * `np.transpose(<array>, axes=[N-1, N-2, ..., 0])` - basically it
-         * is reversing the order of strides and shape.
-         *
-         * This is a fast implementation to handle this special (but very common) case.
-         */
-
-        for (SizeT axis = 0; axis < ndims; axis++) {
-            dst_ndarray->shape[axis] = src_ndarray->shape[ndims - axis - 1];
-            dst_ndarray->strides[axis] = src_ndarray->strides[ndims - axis - 1];
-        }
-    } else {
-        // `np.transpose(<array>, <axes>)`
-
-        // Permute strides and shape according to `axes`, while resolving negative indices in `axes`
-        for (SizeT axis = 0; axis < ndims; axis++) {
-            // `i` cannot be OUT_OF_BOUNDS because of assertions
-            SizeT i = slice::resolve_index_in_length(ndims, axes[axis]);
-
-            dst_ndarray->shape[axis] = src_ndarray->shape[i];
-            dst_ndarray->strides[axis] = src_ndarray->strides[i];
-        }
-    }
-}
-}  // namespace ndarray::transpose
-}  // namespace
-
-extern "C" {
-using namespace ndarray::transpose;
-void __nac3_ndarray_transpose(const NDArray<int32_t>* src_ndarray,
-                              NDArray<int32_t>* dst_ndarray,
-                              int32_t num_axes,
-                              const int32_t* axes) {
-    transpose(src_ndarray, dst_ndarray, num_axes, axes);
-}
-
-void __nac3_ndarray_transpose64(const NDArray<int64_t>* src_ndarray,
-                                NDArray<int64_t>* dst_ndarray,
-                                int64_t num_axes,
-                                const int64_t* axes) {
-    transpose(src_ndarray, dst_ndarray, num_axes, axes);
-}
-}

nac3core/irrt/irrt/numpy/ndarray_basic.hpp

@@ -0,0 +1,221 @@
+#pragma once
+
+#include <irrt/int_defs.hpp>
+#include <irrt/error_context.hpp>
+#include <irrt/numpy/ndarray_def.hpp>
+
+namespace { namespace ndarray { namespace basic {
+    namespace util {
+        // throw an error if there is an axis with negative dimension
+        template <typename SizeT>
+        void assert_shape_no_negative(ErrorContext* errctx, SizeT ndims, const SizeT* shape) {
+            for (SizeT axis = 0; axis < ndims; axis++) {
+                if (shape[axis] < 0) {
+                    errctx->set_error(
+                        errctx->error_ids->value_error,
+                        "negative dimensions are not allowed; axis {0} has dimension {1}",
+                        axis, shape[axis]
+                    );
+                    return;
+                }
+            }
+        }
+
+        // compute the size/# of elements of an ndarray given its shape
+        template <typename SizeT>
+        SizeT calc_size_from_shape(SizeT ndims, const SizeT* shape) {
+            SizeT size = 1;
+            for (SizeT axis = 0; axis < ndims; axis++) size *= shape[axis];
+            return size;
+        }
+
+        // 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/.
+        //
+        // this function might be used in isolation without an ndarray. that's
+        // why it separated out into its own util function.
+        template <typename SizeT>
+        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 axis = ndims - i - 1;
+                dst_strides[axis] = stride_product * itemsize;
+                stride_product *= shape[axis];
+            }
+        }
+
+        template <typename SizeT>
+        void set_indices_by_nth(SizeT ndims, const SizeT* shape, SizeT* indices, SizeT nth) {
+            for (int32_t i = 0; i < ndims; i++) {
+                int32_t axis = ndims - i - 1;
+                int32_t dim = shape[axis];
+
+                indices[axis] = nth % dim;
+                nth /= dim;
+            }
+        }
+    }
+
+    // calculate the size/# of elements of an `ndarray`.
+    // this function corresponds to `np.size(<ndarray>)` or `ndarray.size`
+    template <typename SizeT>
+    SizeT size(NDArray<SizeT>* ndarray) {
+        return util::calc_size_from_shape(ndarray->ndims, ndarray->shape);
+    }
+
+    // calculate the number of bytes of its content of an `ndarray` *in its view*.
+    // this function corresponds to `ndarray.nbytes`
+    template <typename SizeT>
+    SizeT nbytes(NDArray<SizeT>* ndarray) {
+        return size(ndarray) * ndarray->itemsize;
+    }
+
+    template <typename SizeT>
+    void set_strides_by_shape(NDArray<SizeT>* ndarray) {
+        util::set_strides_by_shape(ndarray->itemsize, ndarray->ndims, ndarray->strides, ndarray->shape);
+    }
+
+    template <typename SizeT>
+    uint8_t* get_pelement_by_indices(NDArray<SizeT>* ndarray, const SizeT* indices) {
+        uint8_t* element = ndarray->data;
+        for (SizeT dim_i = 0; dim_i < ndarray->ndims; dim_i++)
+            element += indices[dim_i] * ndarray->strides[dim_i];
+        return element;
+    }
+
+    template <typename SizeT>
+    uint8_t* get_nth_pelement(NDArray<SizeT>* ndarray, SizeT nth) {
+        SizeT* indices = (SizeT*) __builtin_alloca(sizeof(SizeT) * ndarray->ndims);
+        util::set_indices_by_nth(ndarray->ndims, ndarray->shape, indices, nth);
+        return get_pelement_by_indices(ndarray, indices);
+    }
+
+    // get the pointer to the nth element of the ndarray as if it were flattened.
+    template <typename SizeT>
+    uint8_t* checked_get_nth_pelement(NDArray<SizeT>* ndarray, ErrorContext* errctx, SizeT nth) {
+        SizeT arr_size = ndarray->size();
+        if (!(0 <= nth && nth < arr_size)) {
+            errctx->set_error(
+                errctx->error_ids->index_error,
+                "index {0} is out of bounds, valid range is {1} <= index < {2}",
+                nth, 0, arr_size
+            );
+            return 0;
+        }
+        return get_nth_pelement(ndarray, nth);
+    }
+
+    template <typename SizeT>
+    void set_pelement_value(NDArray<SizeT>* ndarray, uint8_t* pelement, const uint8_t* pvalue) {
+        __builtin_memcpy(pelement, pvalue, ndarray->itemsize);
+    }
+
+    template <typename SizeT>
+    void len(ErrorContext* errctx, NDArray<SizeT>* ndarray, SliceIndex* dst_length) {
+        // Error if the ndarray is "unsized" (i.e, ndims == 0)
+        if (ndarray->ndims == 0) {
+            // Error copied from python by doing `len(np.zeros(()))`
+            errctx->set_error(
+                errctx->error_ids->type_error,
+                "len() of unsized object"
+            );
+            return; // Terminate
+        }
+
+        *dst_length = (SliceIndex) ndarray->shape[0];
+    }
+
+    // Copy data from one ndarray to another *OF THE EXACT SAME* ndims, shape, and itemsize.
+    template <typename SizeT>
+    void copy_data(const NDArray<SizeT>* src_ndarray, NDArray<SizeT>* dst_ndarray) {
+        __builtin_assume(src_ndarray->ndims == dst_ndarray->ndims);
+        __builtin_assume(src_ndarray->itemsize == dst_ndarray->itemsize);
+
+        for (SizeT i = 0; i < src_ndarray->size; i++) {
+            auto src_element = ndarray::basic::get_nth_pelement(src_ndarray, i);
+            auto dst_element = ndarray::basic::get_nth_pelement(dst_ndarray, i);
+            ndarray::basic::set_pelement_value(dst_ndarray, dst_element, src_element);
+        }
+    }
+
+    // `copy_data()` with assertions to check ndims, shape, and itemsize between the two ndarrays.
+    template <typename SizeT>
+    void copy_data_checked(ErrorContext* errctx, const NDArray<SizeT>* src_ndarray, NDArray<SizeT>* dst_ndarray) {
+        // NOTE: Out of all error types, runtime error seems appropriate
+
+        // Check ndims
+        if (src_ndarray->ndims != dst_ndarray->ndims) {
+            errctx->set_error(
+                errctx->error_ids->runtime_error,
+                "IRRT copy_data_checked input arrays `ndims` are mismatched"
+            );
+            return; // Terminate
+        }
+
+        // Check shape
+        if (!arrays_match(src_ndarray->ndims, src_ndarray->shape, dst_ndarray->shape)) {
+            errctx->set_error(
+                errctx->error_ids->runtime_error,
+                "IRRT copy_data_checked input arrays `shape` are mismatched"
+            );
+            return; // Terminate
+        }
+
+        // Check itemsize
+        if (src_ndarray->itemsize != dst_ndarray->itemsize) {
+            errctx->set_error(
+                errctx->error_ids->runtime_error,
+                "IRRT copy_data_checked input arrays `itemsize` are mismatched"
+            );
+            return; // Terminate
+        }
+
+        copy_data(src_ndarray, dst_ndarray);
+    }
+} } }
+
+extern "C" {
+    using namespace ndarray::basic;
+
+    uint32_t __nac3_ndarray_size(NDArray<int32_t>* ndarray) {
+        return size(ndarray);
+    }
+
+    uint64_t __nac3_ndarray_size64(NDArray<int64_t>* ndarray) {
+        return size(ndarray);
+    }
+
+    uint32_t __nac3_ndarray_nbytes(NDArray<int32_t>* ndarray) {
+        return nbytes(ndarray);
+    }
+
+    uint64_t __nac3_ndarray_nbytes64(NDArray<int64_t>* ndarray) {
+        return nbytes(ndarray);
+    }
+
+    void __nac3_ndarray_len(ErrorContext* errctx, NDArray<int32_t>* ndarray, SliceIndex* dst_len) {
+        return len(errctx, ndarray, dst_len);
+    }
+
+    void __nac3_ndarray_len64(ErrorContext* errctx, NDArray<int64_t>* ndarray, SliceIndex* dst_len) {
+        return len(errctx, ndarray, dst_len);
+    }
+
+    void __nac3_ndarray_util_assert_shape_no_negative(ErrorContext* errctx, int32_t ndims, int32_t* shape) {
+        util::assert_shape_no_negative(errctx, ndims, shape);
+    }
+
+    void __nac3_ndarray_util_assert_shape_no_negative64(ErrorContext* errctx, int64_t ndims, int64_t* shape) {
+        util::assert_shape_no_negative(errctx, ndims, shape);
+    }
+
+    void __nac3_ndarray_set_strides_by_shape(NDArray<int32_t>* ndarray) {
+        set_strides_by_shape(ndarray);
+    }
+
+    void __nac3_ndarray_set_strides_by_shape64(NDArray<int64_t>* ndarray) {
+        set_strides_by_shape(ndarray);
+    }
+}

nac3core/irrt/irrt/numpy/ndarray_broadcast.hpp

@@ -0,0 +1,135 @@
+#include <irrt/numpy/ndarray_def.hpp>
+
+namespace { namespace ndarray { namespace broadcast {
+    namespace util {
+        template <typename SizeT>
+        void assert_broadcast_shape_to(
+            ErrorContext* errctx,
+            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]) ... ok`
+                - `can_broadcast_shape_to([3], [3, 1]) == false`
+                - `can_broadcast_shape_to([256, 256, 3], [256, 1, 3]) ... ok`
+
+                In cases when the shapes contain zero(es):
+                - `can_broadcast_shape_to([0], [1]) ... ok`
+                - `can_broadcast_shape_to([0], [2]) == false`
+                - `can_broadcast_shape_to([0, 4, 0, 0], [1]) ... ok`
+                - `can_broadcast_shape_to([0, 4, 0, 0], [1, 1, 1, 1]) ... ok`
+                - `can_broadcast_shape_to([0, 4, 0, 0], [1, 4, 1, 1]) ... ok`
+                - `can_broadcast_shape_to([4, 3], [0, 3]) == false`
+                - `can_broadcast_shape_to([4, 3], [0, 0]) == false`
+            */
+
+            // Target ndims must not be smaller than source ndims
+            // e.g., `np.broadcast_to(np.zeros((1, 1, 1, 1)), (1, ))` is prohibited by numpy
+            if (target_ndims < src_ndims) {
+                // Error copied from python by doing the `np.broadcast_to(np.zeros((1, 1, 1, 1)), (1, ))`
+                errctx->set_error(
+                    errctx->error_ids->value_error,
+                    "input operand has more dimensions than allowed by the axis remapping"
+                );
+                return; // Terminate
+            }
+
+            // Implements the rules in https://numpy.org/doc/stable/user/basics.broadcasting.html
+            for (SizeT i = 0; i < src_ndims; i++) {
+                SizeT target_axis = target_ndims - i - 1;
+                SizeT src_axis = src_ndims - i - 1;
+
+                bool target_dim_exists = target_axis >= 0;
+                bool src_dim_exists = src_axis >= 0;
+
+                SizeT target_dim = target_dim_exists ? target_shape[target_axis] : 1;
+                SizeT src_dim = src_dim_exists ? src_shape[src_axis] : 1;
+
+                bool ok = src_dim == 1 || target_dim == src_dim;
+                if (!ok) {
+                    // Error copied from python by doing `np.broadcast_to(np.zeros((3, 1)), (1, 1)),
+                    // but this is the true numpy error:
+                    // "ValueError: operands could not be broadcast together with remapped shapes [original->remapped]: (3,1)  and requested shape (1,1)"
+                    // TODO: we cannot show more than 3 parameters!!
+                    errctx->set_error(
+                        errctx->error_ids->value_error,
+                        "operands could not be broadcast together with remapping shapes [original->remapped]"
+                    );
+                    return; // Terminate
+                }
+            }
+        }
+    }
+
+    // Similar to `np.broadcast_to(<ndarray>, <target_shape>)`
+    // Assumptions:
+    //   - `src_ndarray` 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 `src_ndarray->data`.
+    //   - `dst_ndarray->itemsize` does not have to be set, it will be set to `src_ndarray->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.
+    //   ```
+    template <typename SizeT>
+    void broadcast_to(ErrorContext* errctx, NDArray<SizeT>* src_ndarray, NDArray<SizeT>* dst_ndarray) {
+        dst_ndarray->data = src_ndarray->data;
+        dst_ndarray->itemsize = src_ndarray->itemsize;
+
+        ndarray::broadcast::util::assert_broadcast_shape_to(
+            errctx,
+            dst_ndarray->ndims,
+            dst_ndarray->shape,
+            src_ndarray->ndims,
+            src_ndarray->shape
+        );
+        if (errctx->has_error()) {
+            return; // Propagate error
+        }
+
+        SizeT stride_product = 1;
+        for (SizeT i = 0; i < max(src_ndarray->ndims, dst_ndarray->ndims); i++) {
+            SizeT this_dim_i = src_ndarray->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 && src_ndarray->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 * src_ndarray->itemsize;
+                stride_product *= src_ndarray->shape[this_dim_i]; // NOTE: this_dim_exist must be true here.
+            }
+        }
+    }
+} } }

nac3core/irrt/irrt/numpy/ndarray_def.hpp

@@ -0,0 +1,55 @@
+#pragma once
+
+namespace {
+    // 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.
+        //
+        // This pointer should point to the first element of the ndarray directly
+        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;
+    };
+
+    // Because ndarray is so complicated, its functions are splitted into
+    // different files and namespaces.
+}

nac3core/irrt/irrt/numpy/ndarray_fill.hpp

@@ -0,0 +1,28 @@
+#pragma once
+
+#include <irrt/numpy/ndarray_def.hpp>
+#include <irrt/numpy/ndarray_basic.hpp>
+
+namespace { namespace ndarray { namespace fill {
+    // Fill the ndarray with a value
+    template <typename SizeT>
+    void fill_generic(NDArray<SizeT>* ndarray, const uint8_t* pvalue) {
+        const SizeT size = ndarray::basic::size(ndarray);
+        for (SizeT i = 0; i < size; i++) {
+            uint8_t* pelement = ndarray::basic::get_nth_pelement(ndarray, i); // No need for checked_get_nth_pelement
+            ndarray::basic::set_pelement_value(ndarray, pelement, pvalue);
+        }
+    }
+} } }
+
+extern "C" {
+    using namespace ndarray::fill;
+
+    void __nac3_ndarray_fill_generic(NDArray<int32_t>* ndarray, uint8_t* pvalue) {
+        fill_generic(ndarray, pvalue);
+    }
+
+    void __nac3_ndarray_fill_generic64(NDArray<int64_t>* ndarray, uint8_t* pvalue) {
+        fill_generic(ndarray, pvalue);
+    }
+}

nac3core/irrt/irrt/numpy/ndarray_subscript.hpp

@@ -0,0 +1,169 @@
+#pragma once
+
+#include <irrt/slice.hpp>
+#include <irrt/numpy/ndarray_def.hpp>
+#include <irrt/numpy/ndarray_basic.hpp>
+#include <irrt/error_context.hpp>
+
+namespace {
+    typedef uint8_t NDSubscriptType;
+
+    const NDSubscriptType INPUT_SUBSCRIPT_TYPE_INDEX = 0;
+    const NDSubscriptType INPUT_SUBSCRIPT_TYPE_SLICE = 1;
+
+    struct NDSubscript {
+        // A poor-man's enum variant type
+        NDSubscriptType type;
+
+        /*
+            if type == INPUT_SUBSCRIPT_TYPE_INDEX => `slice` points to a single `SliceIndex`
+            if type == INPUT_SUBSCRIPT_TYPE_SLICE => `slice` points to a single `UserRange`
+
+            `SizeT` is controlled by the caller: `NDSubscript` only cares about where that
+            slice is (the pointer), `NDSubscript` does not care/know about the actual `sizeof()`
+            of the slice value.
+        */
+        uint8_t* data;
+    };
+}
+
+namespace { namespace ndarray { namespace subscript {
+    namespace util {
+        template<typename SizeT>
+        void deduce_ndims_after_slicing(ErrorContext* errctx, SizeT* result, SizeT ndims, SizeT num_ndsubscripts, const NDSubscript* ndsubscripts) {
+            if (num_ndsubscripts > ndims) {
+                // Error copied from python by doing `np.zeros((3, 4))[:, :, :]`
+                errctx->set_error(
+                    errctx->error_ids->index_error,
+                    "too many indices for array: array is {0}-dimensional, but {1} were indexed",
+                    ndims, num_ndsubscripts
+                );
+                return; // Terminate
+            }
+
+            SizeT final_ndims = ndims;
+            for (SizeT i = 0; i < num_ndsubscripts; i++) {
+                if (ndsubscripts[i].type == INPUT_SUBSCRIPT_TYPE_INDEX) {
+                    final_ndims--; // An index demotes the rank by 1
+                }
+            }
+
+            *result = final_ndims;
+        }
+    }
+
+    // To support numpy "basic indexing" https://numpy.org/doc/stable/user/basics.indexing.html#basic-indexing
+    // "Advanced indexing" https://numpy.org/doc/stable/user/basics.indexing.html#advanced-indexing is not supported
+    //
+    // This function supports:
+    //   - "scalar indexing",
+    //   - "slicing and strides",
+    //   - and "dimensional indexing tools" (TODO, but this is really easy to implement).
+    //
+    // 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 `src_ndarray->itemsize`
+    // - `dst_ndarray->shape` and `dst_ndarray.strides` can contain empty values
+    template <typename SizeT>
+    void subscript(ErrorContext* errctx, SliceIndex num_subscripts, NDSubscript* subscripts, NDArray<SizeT>* src_ndarray, 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(src_ndarray->ndims, num_subscripts, subscripts));
+
+        dst_ndarray->data = src_ndarray->data;
+        dst_ndarray->itemsize = src_ndarray->itemsize;
+
+        SizeT src_axis = 0;
+        SizeT dst_axis = 0;
+
+        for (SliceIndex i = 0; i < num_subscripts; i++) {
+            NDSubscript *ndsubscript = &subscripts[i];
+            if (ndsubscript->type == INPUT_SUBSCRIPT_TYPE_INDEX) {
+                // Handle when the ndsubscript is just a single (possibly negative) integer
+                // e.g., `my_array[::2, -5, ::-1]`
+                //                      ^^------ like this
+                SliceIndex input_index = *((SliceIndex*) ndsubscript->data);
+
+                SliceIndex index = slice::resolve_index_in_length(src_ndarray->shape[src_axis], input_index);
+                if (index == slice::OUT_OF_BOUNDS) {
+                    // Error message copied from numpy by doing `np.zeros((3, 4))[100]`
+                    errctx->set_error(
+                        errctx->error_ids->index_error,
+                        "index {0} is out of bounds for axis {1} with size {2}",
+                        input_index, src_axis, src_ndarray->shape[src_axis]
+                    );
+                    return; // Terminate
+                }
+
+                dst_ndarray->data += index * src_ndarray->strides[src_axis]; // Add offset
+
+                // Next
+                src_axis++;
+            } else if (ndsubscript->type == INPUT_SUBSCRIPT_TYPE_SLICE) {
+                // Handle when the ndsubscript is a slice (represented by UserSlice in IRRT)
+                // e.g., `my_array[::2, -5, ::-1]`
+                //                 ^^^------^^^^----- like these
+                UserSlice* input_user_slice = (UserSlice*) ndsubscript->data;
+
+                // TODO: use checked indices
+                Slice slice;
+                input_user_slice->indices_checked(errctx, src_ndarray->shape[src_axis], &slice); // To resolve negative indices and other funny stuff written by the user
+                if (errctx->has_error()) {
+                    return; // Propagate error
+                }
+
+                // 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 += (SizeT) slice.start * src_ndarray->strides[src_axis]; // Add offset (NOTE: no need to `* itemsize`, strides count in # of bytes)
+                dst_ndarray->strides[dst_axis] = ((SizeT) slice.step) * src_ndarray->strides[src_axis]; // Determine stride
+                dst_ndarray->shape[dst_axis] = (SizeT) slice.len(); // Determine shape dimension
+
+                // Next
+                dst_axis++;
+                src_axis++;
+            } else {
+                __builtin_unreachable();
+            }
+        }
+
+        /*
+            Reference python code:
+            ```python
+            dst_ndarray.shape.extend(src_ndarray.shape[src_axis:])
+            dst_ndarray.strides.extend(src_ndarray.strides[src_axis:])
+            ```
+        */
+
+        for (; dst_axis < dst_ndarray->ndims; dst_axis++, src_axis++) {
+            dst_ndarray->shape[dst_axis] = src_ndarray->shape[src_axis];
+            dst_ndarray->strides[dst_axis] = src_ndarray->strides[src_axis];
+        }
+    }
+} } }
+
+extern "C" {
+    using namespace ndarray::subscript;
+
+    void __nac3_ndarray_subscript_deduce_ndims_after_slicing(ErrorContext* errctx, int32_t* result, int32_t ndims, int32_t num_ndsubscripts, const NDSubscript* ndsubscripts) {
+        ndarray::subscript::util::deduce_ndims_after_slicing(errctx, result, ndims, num_ndsubscripts, ndsubscripts);
+    }
+
+    void __nac3_ndarray_subscript_deduce_ndims_after_slicing64(ErrorContext* errctx, int64_t* result, int64_t ndims, int64_t num_ndsubscripts, const NDSubscript* ndsubscripts) {
+        ndarray::subscript::util::deduce_ndims_after_slicing(errctx, result, ndims, num_ndsubscripts, ndsubscripts);
+    }
+
+    void __nac3_ndarray_subscript(ErrorContext* errctx, SliceIndex num_subscripts, NDSubscript* subscripts, NDArray<int32_t>* src_ndarray, NDArray<int32_t> *dst_ndarray) {
+        subscript(errctx, num_subscripts, subscripts, src_ndarray, dst_ndarray);
+    }
+
+    void __nac3_ndarray_subscript64(ErrorContext* errctx, SliceIndex num_subscripts, NDSubscript* subscripts, NDArray<int64_t>* src_ndarray, NDArray<int64_t> *dst_ndarray) {
+        subscript(errctx, num_subscripts, subscripts, src_ndarray, dst_ndarray);
+    }
+}

nac3core/irrt/irrt/range.hpp

@@ -1,47 +0,0 @@
-#pragma once
-
-#include "irrt/debug.hpp"
-#include "irrt/int_types.hpp"
-
-namespace {
-namespace range {
-template<typename T>
-T len(T start, T stop, T step) {
-    // Reference:
-    // https://github.com/python/cpython/blob/9dbd12375561a393eaec4b21ee4ac568a407cdb0/Objects/rangeobject.c#L933
-    if (step > 0 && start < stop)
-        return 1 + (stop - 1 - start) / step;
-    else if (step < 0 && start > stop)
-        return 1 + (start - 1 - stop) / (-step);
-    else
-        return 0;
-}
-}  // namespace range
-
-/**
- * @brief A Python range.
- */
-template<typename T>
-struct Range {
-    T start;
-    T stop;
-    T step;
-
-    /**
-     * @brief Calculate the `len()` of this range.
-     */
-    template<typename SizeT>
-    T len() {
-        debug_assert(SizeT, step != 0);
-        return range::len(start, stop, step);
-    }
-};
-}  // namespace
-
-extern "C" {
-using namespace range;
-
-SliceIndex __nac3_range_slice_len(const SliceIndex start, const SliceIndex end, const SliceIndex step) {
-    return len(start, end, step);
-}
-}

nac3core/irrt/irrt/slice.hpp

@@ -1,156 +1,143 @@
 #pragma once
 
-#include "irrt/debug.hpp"
-#include "irrt/exception.hpp"
-#include "irrt/int_types.hpp"
-#include "irrt/math_util.hpp"
-#include "irrt/range.hpp"
+#include <irrt/int_defs.hpp>
+#include <irrt/slice.hpp>
 
 namespace {
-namespace slice {
-/**
- * @brief Resolve a possibly negative index in a list of a known length.
- *
- * Returns -1 if the resolved index is out of the list's bounds.
- */
-template<typename T>
-T resolve_index_in_length(T length, T index) {
-    T resolved = index < 0 ? length + index : index;
-    if (0 <= resolved && resolved < length) {
-        return resolved;
-    } else {
-        return -1;
-    }
-}
+    struct Slice {
+        SliceIndex start;
+        SliceIndex stop;
+        SliceIndex step;
 
-/**
- * @brief Resolve a slice as a range.
- *
- * This is equivalent to `range(*slice(start, stop, step).indices(length))` in Python.
- */
-template<typename T>
-void indices(bool start_defined,
-             T start,
-             bool stop_defined,
-             T stop,
-             bool step_defined,
-             T step,
-             T length,
-             T* range_start,
-             T* range_stop,
-             T* range_step) {
-    // Reference: https://github.com/python/cpython/blob/main/Objects/sliceobject.c#L388
-    *range_step = step_defined ? step : 1;
-    bool step_is_negative = *range_step < 0;
+        // 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))`
+        SliceIndex len() {
+            SliceIndex 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;
+            }
+        }
+    };
 
-    T lower, upper;
-    if (step_is_negative) {
-        lower = -1;
-        upper = length - 1;
-    } else {
-        lower = 0;
-        upper = length;
-    }
-
-    if (start_defined) {
-        *range_start = start < 0 ? max(lower, start + length) : min(upper, start);
-    } else {
-        *range_start = step_is_negative ? upper : lower;
-    }
-
-    if (stop_defined) {
-        *range_stop = stop < 0 ? max(lower, stop + length) : min(upper, stop);
-    } else {
-        *range_stop = step_is_negative ? lower : upper;
-    }
-}
-}  // namespace slice
-
-/**
- * @brief A Python-like slice with **unresolved** indices.
- */
-template<typename T>
-struct Slice {
-    bool start_defined;
-    T start;
-
-    bool stop_defined;
-    T stop;
-
-    bool step_defined;
-    T step;
-
-    Slice() { this->reset(); }
-
-    void reset() {
-        this->start_defined = false;
-        this->stop_defined = false;
-        this->step_defined = false;
-    }
-
-    void set_start(T start) {
-        this->start_defined = true;
-        this->start = start;
-    }
-
-    void set_stop(T stop) {
-        this->stop_defined = true;
-        this->stop = stop;
-    }
-
-    void set_step(T step) {
-        this->step_defined = true;
-        this->step = step;
-    }
-
-    /**
-     * @brief Resolve this slice as a range.
-     *
-     * In Python, this would be `range(*slice(start, stop, step).indices(length))`.
-     */
-    template<typename SizeT>
-    Range<T> indices(T length) {
-        // Reference:
-        // https://github.com/python/cpython/blob/main/Objects/sliceobject.c#L388
-        debug_assert(SizeT, length >= 0);
-
-        Range<T> result;
-        slice::indices(start_defined, start, stop_defined, stop, step_defined, step, length, &result.start,
-                       &result.stop, &result.step);
-        return result;
-    }
-
-    /**
-     * @brief Like `.indices()` but with assertions.
-     */
-    template<typename SizeT>
-    Range<T> indices_checked(T length) {
-        // TODO: Switch to `SizeT length`
-
-        if (length < 0) {
-            raise_exception(SizeT, EXN_VALUE_ERROR, "length should not be negative, got {0}", length, NO_PARAM,
-                            NO_PARAM);
+    namespace slice {
+        // "Resolve" an index value under a length in Python lists.
+        // If you have a `list` of length 100, `list[-1]` would resolve to `list[100-1] == list[99]`.
+        //
+        // If length == 0, this function returns 0
+        //
+        // If index is out of bounds, this function clamps the value
+        // (to `list[0]` or `list[-1]` in the context of a list and depending on if index is + or -)
+        SliceIndex resolve_index_in_length_clamped(SliceIndex length, SliceIndex index) {
+            if (index < 0) {
+                // Remember that index is negative, so do a plus here
+                return max<SliceIndex>(length + index, 0);
+            } else {
+                return min<SliceIndex>(length, index);
+            }
         }
 
-        if (this->step_defined && this->step == 0) {
-            raise_exception(SizeT, EXN_VALUE_ERROR, "slice step cannot be zero", NO_PARAM, NO_PARAM, NO_PARAM);
+        const SliceIndex OUT_OF_BOUNDS = -1;
+
+        // Like `resolve_index_in_length`.
+        // But also checks if the resolved index is in
+        // bounds (function returns true) or out of bounds
+        // (function returns false); `0 <= resolved index < length` is false).
+        SliceIndex resolve_index_in_length(SliceIndex length, SliceIndex index) {
+            SliceIndex resolved = index < 0 ? length + index : index;
+
+            bool in_bounds = 0 <= resolved && resolved < length;
+            return in_bounds ? resolved : OUT_OF_BOUNDS;
+        }
+    }
+
+    // A user-written Python-like slice.
+    //
+    // i.e., this slice is a triple of either an int or nothing. (e.g., `my_array[:10:2]`, `start` is None)
+    //
+    // You can "resolve" a `UserSlice` by using `user_slice.indices(<length>)`
+    struct UserSlice {
+        // Did the user specify `start`? If 0, `start` is undefined (and contains an empty value)
+        bool start_defined;
+        SliceIndex start;
+
+        // Similar to `start_defined`
+        bool stop_defined;
+        SliceIndex stop;
+
+        // Similar to `start_defined`
+        bool step_defined;
+        SliceIndex step;
+
+        // Convenient constructor for C++ internal use only (say testing)
+        UserSlice() {
+            this->reset();
         }
 
-        return this->indices<SizeT>(length);
-    }
-};
-}  // namespace
+        void reset() {
+            this->start_defined = false;
+            this->stop_defined = false;
+            this->step_defined = false;
+        }
 
-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;
-}
-}
+        void set_start(SliceIndex start) {
+            this->start_defined = true;
+            this->start = start;
+        }
+
+        void set_stop(SliceIndex stop) {
+            this->stop_defined = true;
+            this->stop = stop;
+        }
+
+        void set_step(SliceIndex step) {
+            this->step_defined = true;
+            this->step = step;
+        }
+
+        // Like Python's `slice(start, stop, step).indices(length)`
+        void indices(SliceIndex length, Slice* result) {
+            // NOTE: This function implements Python's `slice.indices` *FAITHFULLY*.
+            // SEE: https://github.com/python/cpython/blob/f62161837e68c1c77961435f1b954412dd5c2b65/Objects/sliceobject.c#L546
+            result->step = step_defined ? step : 1;
+            bool step_is_negative = result->step < 0;
+
+            if (start_defined) {
+                result->start = slice::resolve_index_in_length_clamped(length, start);
+            } else {
+                result->start = step_is_negative ? length - 1 : 0;
+            }
+            
+            if (stop_defined) {
+                result->stop = slice::resolve_index_in_length_clamped(length, stop);
+            } else {
+                result->stop = step_is_negative ? -1 : length;
+            }
+        }
+
+        // `indices()` but asserts `this->step != 0` and `this->length >= 0`
+        void indices_checked(ErrorContext* errctx, SliceIndex length, Slice* result) {
+            if (length < 0) {
+                errctx->set_error(
+                    errctx->error_ids->value_error,
+                    "length should not be negative, got {0}", // Edited. Error message copied from python by doing `slice(0, 0, 0).indices(100)`
+                    length
+                );
+                return;
+            }
+
+            if (this->step_defined && this->step == 0) {
+                // Error message
+                errctx->set_error(
+                    errctx->error_ids->value_error,
+                    "slice step cannot be zero" // Error message copied from python by doing `slice(0, 0, 0).indices(100)`
+                );
+                return;
+            }
+            this->indices(length, result);
+        }
+    };
+}

nac3core/irrt/irrt/string.hpp

@@ -1,23 +0,0 @@
-#pragma once
-
-#include "irrt/int_types.hpp"
-
-namespace {
-template<typename SizeT>
-bool __nac3_str_eq_impl(const char* str1, SizeT len1, const char* str2, SizeT len2) {
-    if (len1 != len2) {
-        return 0;
-    }
-    return __builtin_memcmp(str1, str2, static_cast<SizeT>(len1)) == 0;
-}
-}  // namespace
-
-extern "C" {
-bool nac3_str_eq(const char* str1, uint32_t len1, const char* str2, uint32_t len2) {
-    return __nac3_str_eq_impl<uint32_t>(str1, len1, str2, len2);
-}
-
-bool nac3_str_eq64(const char* str1, uint64_t len1, const char* str2, uint64_t len2) {
-    return __nac3_str_eq_impl<uint64_t>(str1, len1, str2, len2);
-}
-}

nac3core/irrt/irrt/utils.hpp

@@ -0,0 +1,78 @@
+#pragma once
+
+#include <irrt/int_defs.hpp>
+
+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;
+    }
+
+    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;
+    }
+
+    namespace cstr_utils {
+        bool is_empty(const char* str) {
+            return str[0] == '\0';
+        }
+
+        int8_t compare(const char* a, const char* b) {
+            uint32_t i = 0;
+            while (true) {
+                if (a[i] < b[i]) {
+                    return -1;
+                } else if (a[i] > b[i]) {
+                    return 1;
+                } else { // a[i] == b[i]
+                    if (a[i] == '\0') {
+                        return 0;
+                    } else {
+                        i++;
+                    }
+                }
+            }
+        }
+
+        int8_t equal(const char* a, const char* b) {
+            return compare(a, b) == 0;
+        }
+
+        uint32_t length(const char* str) {
+            uint32_t length = 0;
+            while (*str != '\0') {
+                length++;
+                str++;
+            }
+            return length;
+        }
+
+        bool copy(const char* src, char* dst, uint32_t dst_max_size) {
+            for (uint32_t i = 0; i < dst_max_size; i++) {
+                bool is_last = i + 1 == dst_max_size;
+                if (is_last && src[i] != '\0') {
+                    dst[i] = '\0';
+                    return false;
+                }
+
+                if (src[i] == '\0') {
+                    dst[i] = '\0';
+                    return true;
+                }
+
+                dst[i] = src[i];
+            }
+
+            __builtin_unreachable();
+        }
+    }
+}

nac3core/irrt/irrt_everything.hpp

@@ -0,0 +1,12 @@
+#pragma once
+
+#include <irrt/core.hpp>
+#include <irrt/error_context.hpp>
+#include <irrt/int_defs.hpp>
+#include <irrt/numpy/ndarray_basic.hpp>
+#include <irrt/numpy/ndarray_broadcast.hpp>
+#include <irrt/numpy/ndarray_def.hpp>
+#include <irrt/numpy/ndarray_fill.hpp>
+#include <irrt/numpy/ndarray_subscript.hpp>
+#include <irrt/slice.hpp>
+#include <irrt/utils.hpp>

nac3core/irrt/irrt_test.cpp

@@ -0,0 +1,25 @@
+// 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>
+
+#include <irrt_everything.hpp>
+
+#include <test/core.hpp>
+#include <test/test_core.hpp>
+#include <test/test_ndarray_basic.hpp>
+#include <test/test_ndarray_subscript.hpp>
+#include <test/test_ndarray_broadcast.hpp>
+#include <test/test_slice.hpp>
+
+int main() {
+    // Be wise about the order of suites!!
+    test::core::run();
+    test::slice::run();
+    test::ndarray_basic::run();
+    test::ndarray_subscript::run();
+    test::ndarray_broadcast::run();
+    return 0;
+}

nac3core/irrt/test/core.hpp

@@ -0,0 +1,143 @@
+#pragma once
+
+// Include this header for every test_*.cpp
+
+#include <cstdint>
+#include <cstdio>
+#include <cstdlib>
+
+#include <test/print.hpp>
+
+// Some utils can be used here
+#include <irrt/utils.hpp>
+
+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__)
+
+void test_fail() {
+    printf("[!] Test failed. Exiting with status code 1.\n");
+    exit(1);
+}
+
+void print_assertion_passed(const char* file, int line) {
+    printf("[*] Assertion passed on %s:%d\n", file, line);
+}
+
+void print_assertion_failed(const char* file, int line) {
+    printf("[!] Assertion failed on %s:%d\n", file, line);
+}
+
+void __assert_true(const char* file, int line, bool cond) {
+    if (cond) {
+        print_assertion_passed(file, line);
+    } else {
+        print_assertion_failed(file, line);
+        test_fail();
+    }
+}
+
+#define assert_true(cond) __assert_true(__FILE__, __LINE__, cond)
+
+template <typename T>
+void __assert_arrays_match(const char* file, int line, int len, const T* expected, const T* got) {
+    if (arrays_match(len, expected, got)) {
+        print_assertion_passed(file, line);
+    } else {
+        print_assertion_failed(file, line);
+        printf("Expect = ");
+        print_array(len, expected);
+        printf("\n");
+        printf("   Got = ");
+        print_array(len, got);
+        printf("\n");
+        test_fail();
+    }
+}
+
+#define assert_arrays_match(len, expected, got) __assert_arrays_match(__FILE__, __LINE__, len, expected, got)
+
+template <typename T>
+void __assert_values_match(const char* file, int line, T expected, T got) {
+    if (expected == got) {
+        print_assertion_passed(file, line);
+    } else {
+        print_assertion_failed(file, line);
+        printf("Expect = ");
+        print_value(expected);
+        printf("\n");
+        printf("   Got = ");
+        print_value(got);
+        printf("\n");
+        test_fail();
+    }
+}
+
+#define assert_values_match(expected, got) __assert_values_match(__FILE__, __LINE__, expected, got)
+
+// A fake set of ErrorIds for testing only
+const ErrorIds TEST_ERROR_IDS = {
+    .index_error = 0,
+    .value_error = 1,
+    .assertion_error = 2,
+    .runtime_error = 3,
+};
+
+ErrorContext create_testing_errctx() {
+    // Everything is global so it is fine to directly return a struct ErrorContext
+    ErrorContext errctx;
+    errctx.initialize(&TEST_ERROR_IDS);
+    return errctx;
+}
+
+void print_errctx_content(ErrorContext* errctx) {
+    if (errctx->has_error()) {
+        printf(
+            "(Error ID %d): %s ... where param1 = %ld, param2 = %ld, param3 = %ld\n",
+            errctx->error_id,
+            errctx->message_template,
+            errctx->param1,
+            errctx->param2,
+            errctx->param3
+        );
+    } else {
+        printf("<no error>\n");
+    }
+}
+
+void __assert_errctx_no_error(const char* file, int line, ErrorContext* errctx) {
+    if (errctx->has_error()) {
+        print_assertion_failed(file, line);
+        printf("Expecting no error but caught the following:\n\n");
+        print_errctx_content(errctx);
+        test_fail();
+    }
+}
+
+#define assert_errctx_no_error(errctx) __assert_errctx_no_error(__FILE__, __LINE__, errctx)
+
+void __assert_errctx_has_error(const char* file, int line, ErrorContext* errctx, ErrorId expected_error_id) {
+    if (errctx->has_error()) {
+        if (errctx->error_id == expected_error_id) {
+            // OK
+        } else {
+            // Otherwise it got the wrong kind of error
+            print_assertion_failed(file, line);
+            printf(
+                "Expecting error id %d but got error id %d. Error caught:\n\n",
+                expected_error_id,
+                errctx->error_id
+            );
+            print_errctx_content(errctx);
+            test_fail();
+        }
+    } else {
+        print_assertion_failed(file, line);
+        printf("Expecting an error, but there is none.");
+        test_fail();
+    }
+}
+
+#define assert_errctx_has_error(errctx, expected_error_id) __assert_errctx_has_error(__FILE__, __LINE__, errctx, expected_error_id)

nac3core/irrt/test/print.hpp

@@ -0,0 +1,118 @@
+#pragma once
+
+#include <cstdlib>
+#include <cstdio>
+
+template <class T>
+void print_value(const T& value);
+
+template <>
+void print_value(const bool& value) {
+    printf("%s", value ? "true" : "false");
+}
+
+template <>
+void print_value(const int8_t& value) {
+    printf("%d", value);
+}
+
+template <>
+void print_value(const int32_t& value) {
+    printf("%d", value);
+}
+
+template <>
+void print_value(const uint8_t& value) {
+    printf("%u", value);
+}
+
+template <>
+void print_value(const uint32_t& value) {
+    printf("%u", value);
+}
+
+template <>
+void print_value(const float& value) {
+    printf("%f", value);
+}
+
+template <>
+void print_value(const double& value) {
+    printf("%f", value);
+}
+
+void print_repeated(const char *str, int count) {
+    for (int i = 0; i < count; i++) {
+        printf("%s", str);
+    }
+}
+
+template <typename T>
+void print_array(int len, const T* as) {
+    printf("[");
+    for (int i = 0; i < len; i++) {
+        if (i != 0) printf(", ");
+        print_value(as[i]);
+    }
+    printf("]");
+}
+
+template<typename ElementT, typename SizeT>
+void __print_ndarray_aux(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::basic::util::set_indices_by_nth(ndarray->ndims, ndarray->shape, indices, *cursor);
+            ElementT* pelement = (ElementT*) ndarray::basic::get_pelement_by_indices<SizeT>(ndarray, indices);
+            ElementT element = *pelement;
+
+            if (i != 0) printf(", "); // List delimiter
+            print_value(element);
+            printf("(@");
+            print_array(ndarray->ndims, indices);
+            printf(")");
+
+            (*cursor)++;
+        }
+        printf("]");
+    } else {
+        printf("[");
+        for (SizeT i = 0; i < ndarray->shape[depth]; i++) {
+            __print_ndarray_aux<ElementT, SizeT>(
+                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 ElementT, typename SizeT>
+void print_ndarray(NDArray<SizeT>* ndarray) {
+    if (ndarray->ndims == 0) {
+        printf("<empty ndarray>");
+    } else {
+        SizeT cursor = 0;
+        __print_ndarray_aux<ElementT, SizeT>(true, true, &cursor, 0, ndarray);
+    }
+    printf("\n");
+}

nac3core/irrt/test/test_core.hpp

@@ -0,0 +1,19 @@
+#pragma once
+
+#include <test/core.hpp>
+#include <irrt_everything.hpp>
+
+namespace test {
+namespace core {
+void test_int_exp() {
+    BEGIN_TEST();
+
+    assert_values_match(125, __nac3_int_exp_impl<int32_t>(5, 3));
+    assert_values_match(3125, __nac3_int_exp_impl<int32_t>(5, 5));
+}
+
+void run() {
+    test_int_exp();
+}
+}
+}

nac3core/irrt/test/test_ndarray_basic.hpp

@@ -0,0 +1,47 @@
+#pragma once
+
+#include <test/core.hpp>
+#include <irrt_everything.hpp>
+
+namespace test {
+namespace ndarray_basic {
+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(210, ndarray::basic::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(0, ndarray::basic::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::basic::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(4, expected_strides, strides);
+}
+
+void run() {
+    test_calc_size_from_shape_normal();
+    test_calc_size_from_shape_has_zero();
+    test_set_strides_by_shape();
+}
+}
+}

nac3core/irrt/test/test_ndarray_broadcast.hpp

@@ -0,0 +1,72 @@
+#pragma once
+
+#include <test/core.hpp>
+#include <irrt_everything.hpp>
+
+namespace test { namespace ndarray_broadcast {
+    void test_ndarray_broadcast_1() {
+        /*
+            ```python
+            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]]]
+            
+            assert array.strides == (0, 0, 8)
+            # and then pick some values in `array` and check them...
+            ```
+        */
+        BEGIN_TEST();
+
+        // Prepare src_ndarray
+        double src_data[4] = { 19.9, 29.9, 39.9, 49.9 };
+        const int32_t src_ndims = 2;
+        int32_t src_shape[src_ndims] = {1, 4};
+        int32_t src_strides[src_ndims] = {};
+        NDArray<int32_t> src_ndarray = {
+            .data = (uint8_t*) src_data,
+            .itemsize = sizeof(double),
+            .ndims = src_ndims,
+            .shape = src_shape,
+            .strides = src_strides
+        };
+        ndarray::basic::set_strides_by_shape(&src_ndarray);
+
+        // Prepare dst_ndarray
+        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
+        };
+
+        // Broadcast
+        ErrorContext errctx = create_testing_errctx();
+        ndarray::broadcast::broadcast_to(&errctx, &src_ndarray, &dst_ndarray);
+        assert_errctx_no_error(&errctx);
+
+        assert_arrays_match(dst_ndims, ((int32_t[]) { 0, 0, 8 }), dst_ndarray.strides);
+
+        assert_values_match(19.9, *((double*) ndarray::basic::get_pelement_by_indices(&dst_ndarray, ((int32_t[]) {0, 0, 0}))));
+        assert_values_match(29.9, *((double*) ndarray::basic::get_pelement_by_indices(&dst_ndarray, ((int32_t[]) {0, 0, 1}))));
+        assert_values_match(39.9, *((double*) ndarray::basic::get_pelement_by_indices(&dst_ndarray, ((int32_t[]) {0, 0, 2}))));
+        assert_values_match(49.9, *((double*) ndarray::basic::get_pelement_by_indices(&dst_ndarray, ((int32_t[]) {0, 0, 3}))));
+        assert_values_match(19.9, *((double*) ndarray::basic::get_pelement_by_indices(&dst_ndarray, ((int32_t[]) {0, 1, 0}))));
+        assert_values_match(29.9, *((double*) ndarray::basic::get_pelement_by_indices(&dst_ndarray, ((int32_t[]) {0, 1, 1}))));
+        assert_values_match(39.9, *((double*) ndarray::basic::get_pelement_by_indices(&dst_ndarray, ((int32_t[]) {0, 1, 2}))));
+        assert_values_match(49.9, *((double*) ndarray::basic::get_pelement_by_indices(&dst_ndarray, ((int32_t[]) {0, 1, 3}))));
+        assert_values_match(49.9, *((double*) ndarray::basic::get_pelement_by_indices(&dst_ndarray, ((int32_t[]) {1, 2, 3}))));
+    }
+
+    void run() {
+        test_ndarray_broadcast_1();
+    }
+}}

nac3core/irrt/test/test_ndarray_subscript.hpp

@@ -0,0 +1,233 @@
+#pragma once
+
+#include <test/core.hpp>
+#include <irrt_everything.hpp>
+
+namespace test { namespace ndarray_subscript {
+    void test_ndsubscript_normal_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();
+
+        // Prepare src_ndarray
+        double src_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 src_itemsize = sizeof(double);
+        const int32_t src_ndims = 2;
+        int32_t src_shape[src_ndims] = { 3, 4 };
+        int32_t src_strides[src_ndims] = {};
+        NDArray<int32_t> src_ndarray = {
+            .data = (uint8_t*) src_data,
+            .itemsize = src_itemsize,
+            .ndims = src_ndims,
+            .shape = src_shape,
+            .strides = src_strides
+        };
+        ndarray::basic::set_strides_by_shape(&src_ndarray);
+
+        // Prepare dst_ndarray
+        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 subscripts in `ndarray[-2::, 1::2]`
+        UserSlice subscript_1;
+        subscript_1.set_start(-2);
+
+        UserSlice subscript_2;
+        subscript_2.set_start(1);
+        subscript_2.set_step(2);
+
+        const int32_t num_ndsubscripts = 2;
+        NDSubscript ndsubscripts[num_ndsubscripts] = {
+            { .type = INPUT_SUBSCRIPT_TYPE_SLICE, .data = (uint8_t*) &subscript_1 },
+            { .type = INPUT_SUBSCRIPT_TYPE_SLICE, .data = (uint8_t*) &subscript_2 }
+        };
+
+        ErrorContext errctx = create_testing_errctx();
+        ndarray::subscript::subscript(&errctx, num_ndsubscripts, ndsubscripts, &src_ndarray, &dst_ndarray);
+        assert_errctx_no_error(&errctx);
+
+        int32_t expected_shape[dst_ndims] = { 2, 2 };
+        int32_t expected_strides[dst_ndims] = { 32, 16 };
+
+        assert_arrays_match(dst_ndims, expected_shape, dst_ndarray.shape);
+        assert_arrays_match(dst_ndims, expected_strides, dst_ndarray.strides);
+
+        // dst_ndarray[0, 0]
+        assert_values_match(
+            5.0,
+            *((double *) ndarray::basic::get_pelement_by_indices(&dst_ndarray, (int32_t[dst_ndims]) { 0, 0 }))
+        );
+        // dst_ndarray[0, 1]
+        assert_values_match(
+            7.0,
+            *((double *) ndarray::basic::get_pelement_by_indices(&dst_ndarray, (int32_t[dst_ndims]) { 0, 1 }))
+        );
+        // dst_ndarray[1, 0]
+        assert_values_match(
+            9.0,
+            *((double *) ndarray::basic::get_pelement_by_indices(&dst_ndarray, (int32_t[dst_ndims]) { 1, 0 }))
+        );
+        // dst_ndarray[1, 1]
+        assert_values_match(
+            11.0,
+            *((double *) ndarray::basic::get_pelement_by_indices(&dst_ndarray, (int32_t[dst_ndims]) { 1, 1 }))
+        );
+    }
+
+    void test_ndsubscript_normal_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
+            ```
+        */
+        BEGIN_TEST();
+
+        // Prepare src_ndarray
+        double src_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 src_itemsize = sizeof(double);
+        const int32_t src_ndims = 2;
+        int32_t src_shape[src_ndims] = { 3, 4 };
+        int32_t src_strides[src_ndims] = {};
+        NDArray<int32_t> src_ndarray = {
+            .data = (uint8_t*) src_data,
+            .itemsize = src_itemsize,
+            .ndims = src_ndims,
+            .shape = src_shape,
+            .strides = src_strides
+        };
+        ndarray::basic::set_strides_by_shape(&src_ndarray);
+
+        // Prepare dst_ndarray
+        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 subscripts in `ndarray[2, ::-2]`
+        int32_t subscript_1 = 2;
+
+        UserSlice subscript_2;
+        subscript_2.set_step(-2);
+
+        const int32_t num_ndsubscripts = 2;
+        NDSubscript ndsubscripts[num_ndsubscripts] = {
+            { .type = INPUT_SUBSCRIPT_TYPE_INDEX, .data = (uint8_t*) &subscript_1 },
+            { .type = INPUT_SUBSCRIPT_TYPE_SLICE, .data = (uint8_t*) &subscript_2 }
+        };
+
+        ErrorContext errctx = create_testing_errctx();
+        ndarray::subscript::subscript(&errctx, num_ndsubscripts, ndsubscripts, &src_ndarray, &dst_ndarray);
+        assert_errctx_no_error(&errctx);
+
+        int32_t expected_shape[dst_ndims] = { 2 };
+        int32_t expected_strides[dst_ndims] = { -16 };
+        assert_arrays_match(dst_ndims, expected_shape, dst_ndarray.shape);
+        assert_arrays_match(dst_ndims, expected_strides, dst_ndarray.strides);
+
+        assert_values_match(
+            11.0,
+            *((double *) ndarray::basic::get_pelement_by_indices(&dst_ndarray, (int32_t[dst_ndims]) { 0 }))
+        );
+        assert_values_match(
+            9.0,
+            *((double *) ndarray::basic::get_pelement_by_indices(&dst_ndarray, (int32_t[dst_ndims]) { 1 }))
+        );
+    }
+
+    void test_ndsubscript_index_subscript_out_of_bounds() {
+        /*
+            # Consider `my_array`
+
+            print(my_array.shape)
+            # (4, 5, 6)
+
+            my_array[2, 100] # error, index subscript at axis 1 is out of bounds
+        */
+        BEGIN_TEST();
+
+        // Prepare src_ndarray
+        const int32_t src_ndims = 2;
+        int32_t src_shape[src_ndims] = { 3, 4 };
+        int32_t src_strides[src_ndims] = {};
+        NDArray<int32_t> src_ndarray = {
+            .data = (uint8_t*) nullptr, // placeholder, we wouldn't access it
+            .itemsize = sizeof(double), // placeholder
+            .ndims = src_ndims,
+            .shape = src_shape,
+            .strides = src_strides
+        };
+        ndarray::basic::set_strides_by_shape(&src_ndarray);
+
+        // Create the subscripts in `my_array[2, 100]`
+        int32_t subscript_1 = 2;
+        int32_t subscript_2 = 100;
+
+        const int32_t num_ndsubscripts = 2;
+        NDSubscript ndsubscripts[num_ndsubscripts] = {
+            { .type = INPUT_SUBSCRIPT_TYPE_INDEX, .data = (uint8_t*) &subscript_1 },
+            { .type = INPUT_SUBSCRIPT_TYPE_INDEX, .data = (uint8_t*) &subscript_2 }
+        };
+
+        // Prepare dst_ndarray
+        const int32_t dst_ndims = 0;
+        int32_t dst_shape[dst_ndims] = {};
+        int32_t dst_strides[dst_ndims] = {};
+        NDArray<int32_t> dst_ndarray = {
+            .data = nullptr, // placehloder
+            .ndims = dst_ndims,
+            .shape = dst_shape,
+            .strides = dst_strides
+        };
+
+        ErrorContext errctx = create_testing_errctx();
+        ndarray::subscript::subscript(&errctx, num_ndsubscripts, ndsubscripts, &src_ndarray, &dst_ndarray);
+        assert_errctx_has_error(&errctx, errctx.error_ids->index_error);
+    }
+
+    void run() {
+        test_ndsubscript_normal_1();
+        test_ndsubscript_normal_2();
+        test_ndsubscript_index_subscript_out_of_bounds();
+    }
+} }

nac3core/irrt/test/test_slice.hpp

@@ -0,0 +1,96 @@
+#pragma once
+
+#include <test/core.hpp>
+#include <irrt_everything.hpp>
+
+namespace test {
+namespace slice {
+void test_slice_normal() {
+    // Normal situation
+    BEGIN_TEST();
+
+    UserSlice user_slice;
+    user_slice.set_stop(5);
+
+    Slice slice;
+    user_slice.indices(100, &slice);
+
+    printf("%d, %d, %d\n", slice.start, slice.stop, slice.step);
+
+    assert_values_match(0, slice.start);
+    assert_values_match(5, slice.stop);
+    assert_values_match(1, slice.step);
+}
+
+void test_slice_start_too_large() {
+    // Start is too large and should be clamped to length
+    BEGIN_TEST();
+
+    UserSlice user_slice;
+    user_slice.set_start(400);
+
+    Slice slice;
+    user_slice.indices(100, &slice);
+
+    assert_values_match(100, slice.start);
+    assert_values_match(100, slice.stop);
+    assert_values_match(1, slice.step);
+}
+
+void test_slice_negative_start_stop() {
+    // Negative start/stop should be resolved
+    BEGIN_TEST();
+
+    UserSlice user_slice;
+    user_slice.set_start(-10);
+    user_slice.set_stop(-5);
+
+    Slice slice;
+    user_slice.indices(100, &slice);
+
+    assert_values_match(90, slice.start);
+    assert_values_match(95, slice.stop);
+    assert_values_match(1, slice.step);
+}
+
+void test_slice_only_negative_step() {
+    // Things like `[::-5]` should be handled correctly
+    BEGIN_TEST();
+
+    UserSlice user_slice;
+    user_slice.set_step(-5);
+
+    Slice slice;
+    user_slice.indices(100, &slice);
+
+    assert_values_match(99, slice.start);
+    assert_values_match(-1, slice.stop);
+    assert_values_match(-5, slice.step);
+}
+
+void test_slice_step_zero() {
+    // Step = 0 is a value error
+    BEGIN_TEST();
+
+    ErrorContext errctx = create_testing_errctx();
+
+    UserSlice user_slice;
+    user_slice.set_start(2);
+    user_slice.set_stop(12);
+    user_slice.set_step(0);
+
+    Slice slice;
+    user_slice.indices_checked(&errctx, 100, &slice);
+
+    assert_errctx_has_error(&errctx, errctx.error_ids->value_error);
+}
+
+void run() {
+    test_slice_normal();
+    test_slice_start_too_large();
+    test_slice_negative_start_stop();
+    test_slice_only_negative_step();
+    test_slice_step_zero();
+}
+}
+}

nac3core/nac3core_derive/Cargo.toml

@@ -1,21 +0,0 @@
-[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

@@ -1,320 +0,0 @@
-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

@@ -1,9 +0,0 @@
-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

@@ -1,20 +0,0 @@
-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

@@ -1,18 +0,0 @@
-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

@@ -1,18 +0,0 @@
-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

@@ -1,18 +0,0 @@
-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

@@ -1,18 +0,0 @@
-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

@@ -1,10 +0,0 @@
-#[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

@@ -1,9 +1,3 @@
-use std::collections::HashMap;
-
-use indexmap::IndexMap;
-
-use nac3parser::ast::StrRef;
-
 use crate::{
     symbol_resolver::SymbolValue,
     toplevel::DefinitionId,
@@ -15,6 +9,10 @@ use crate::{
     },
 };
 
+use indexmap::IndexMap;
+use nac3parser::ast::StrRef;
+use std::collections::HashMap;
+
 pub struct ConcreteTypeStore {
     store: Vec<ConcreteTypeEnum>,
 }
@@ -27,7 +25,6 @@ pub struct ConcreteFuncArg {
     pub name: StrRef,
     pub ty: ConcreteType,
     pub default_value: Option<SymbolValue>,
-    pub is_vararg: bool,
 }
 
 #[derive(Clone, Debug)]
@@ -49,17 +46,12 @@ pub enum ConcreteTypeEnum {
     TPrimitive(Primitive),
     TTuple {
         ty: Vec<ConcreteType>,
-        is_vararg_ctx: bool,
     },
     TObj {
         obj_id: DefinitionId,
         fields: HashMap<StrRef, (ConcreteType, bool)>,
         params: IndexMap<TypeVarId, ConcreteType>,
     },
-    TModule {
-        module_id: DefinitionId,
-        methods: HashMap<StrRef, (ConcreteType, bool)>,
-    },
     TVirtual {
         ty: ConcreteType,
     },
@@ -110,16 +102,8 @@ impl ConcreteTypeStore {
                 .iter()
                 .map(|arg| ConcreteFuncArg {
                     name: arg.name,
-                    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)
-                    },
+                    ty: 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),
@@ -174,12 +158,11 @@ impl ConcreteTypeStore {
             cache.insert(ty, None);
             let ty_enum = unifier.get_ty(ty);
             let result = match &*ty_enum {
-                TypeEnum::TTuple { ty, is_vararg_ctx } => ConcreteTypeEnum::TTuple {
+                TypeEnum::TTuple { ty } => 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,
@@ -209,19 +192,6 @@ impl ConcreteTypeStore {
                         })
                         .collect(),
                 },
-                TypeEnum::TModule { module_id, attributes } => ConcreteTypeEnum::TModule {
-                    module_id: *module_id,
-                    methods: attributes
-                        .iter()
-                        .filter_map(|(name, ty)| match &*unifier.get_ty(ty.0) {
-                            TypeEnum::TFunc(..) | TypeEnum::TObj { .. } => None,
-                            _ => Some((
-                                *name,
-                                (self.from_unifier_type(unifier, primitives, ty.0, cache), ty.1),
-                            )),
-                        })
-                        .collect(),
-                },
                 TypeEnum::TVirtual { ty } => ConcreteTypeEnum::TVirtual {
                     ty: self.from_unifier_type(unifier, primitives, *ty, cache),
                 },
@@ -278,12 +248,11 @@ impl ConcreteTypeStore {
                 *cache.get_mut(&cty).unwrap() = Some(ty);
                 return ty;
             }
-            ConcreteTypeEnum::TTuple { ty, is_vararg_ctx } => TypeEnum::TTuple {
+            ConcreteTypeEnum::TTuple { ty } => 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) }
@@ -301,15 +270,6 @@ impl ConcreteTypeStore {
                     TypeVar { id, ty }
                 })),
             },
-            ConcreteTypeEnum::TModule { module_id, methods } => TypeEnum::TModule {
-                module_id: *module_id,
-                attributes: methods
-                    .iter()
-                    .map(|(name, cty)| {
-                        (*name, (self.to_unifier_type(unifier, primitives, cty.0, cache), cty.1))
-                    })
-                    .collect::<HashMap<_, _>>(),
-            },
             ConcreteTypeEnum::TFunc { args, ret, vars } => TypeEnum::TFunc(FunSignature {
                 args: args
                     .iter()
@@ -317,7 +277,6 @@ 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

@@ -1,10 +1,8 @@
-use inkwell::{
-    attributes::{Attribute, AttributeLoc},
-    values::{BasicValueEnum, CallSiteValue, FloatValue, IntValue},
-};
+use inkwell::attributes::{Attribute, AttributeLoc};
+use inkwell::values::{BasicValueEnum, CallSiteValue, FloatValue, IntValue};
 use itertools::Either;
 
-use super::CodeGenContext;
+use crate::codegen::CodeGenContext;
 
 /// Macro to generate extern function
 /// Both function return type and function parameter type are `FloatValue`
@@ -15,11 +13,11 @@ use super::CodeGenContext;
 /// * `$extern_fn:literal`: Name of underlying extern function
 ///
 /// Optional Arguments:
-/// * `$(,$attributes:literal)*)`: Attributes linked with the extern function.
-///   The default attributes are "mustprogress", "nofree", "nounwind", "willreturn", and "writeonly".
-///   These will be used unless other attributes are specified
+/// * `$(,$attributes:literal)*)`: Attributes linked with the extern function
+/// The default attributes are "mustprogress", "nofree", "nounwind", "willreturn", and "writeonly"
+/// These will be used unless other attributes are specified
 /// * `$(,$args:ident)*`: Operands of the extern function
-///   The data type of these operands will be set to `FloatValue`
+/// The data type of these operands will be set to `FloatValue`
 ///  
 macro_rules! generate_extern_fn {
     ("unary", $fn_name:ident, $extern_fn:literal) => {
@@ -132,62 +130,3 @@ pub fn call_ldexp<'ctx>(
         .map(Either::unwrap_left)
         .unwrap()
 }
-
-/// Macro to generate `np_linalg` and `sp_linalg` functions
-/// The function takes as input `NDArray` and returns ()
-///
-/// Arguments:
-/// * `$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

@@ -1,27 +1,20 @@
-use inkwell::{
-    context::Context,
-    targets::TargetMachine,
-    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::{
+    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};
 
 pub trait CodeGenerator {
     /// Return the module name for the code generator.
     fn get_name(&self) -> &str;
 
-    /// Return an instance of [`IntType`] corresponding to the type of `size_t` for this instance.
-    ///
-    /// Prefer using [`CodeGenContext::get_size_type`] if [`CodeGenContext`] is available, as it is
-    /// equivalent to this function in a more concise syntax.
     fn get_size_type<'ctx>(&self, ctx: &'ctx Context) -> IntType<'ctx>;
 
     /// Generate function call and returns the function return value.
@@ -64,7 +57,6 @@ 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>(
@@ -139,39 +131,6 @@ pub trait CodeGenerator {
         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.
     /// Return true if the while loop must early return
     fn gen_while(
@@ -274,27 +233,19 @@ pub struct DefaultCodeGenerator {
 
 impl DefaultCodeGenerator {
     #[must_use]
-    pub fn new(name: String, size_t: IntType<'_>) -> DefaultCodeGenerator {
-        assert!(matches!(size_t.get_bit_width(), 32 | 64));
-        DefaultCodeGenerator { name, size_t: size_t.get_bit_width() }
-    }
-
-    #[must_use]
-    pub fn with_target_machine(
-        name: String,
-        ctx: &Context,
-        target_machine: &TargetMachine,
-    ) -> DefaultCodeGenerator {
-        let llvm_usize = ctx.ptr_sized_int_type(&target_machine.get_target_data(), None);
-        Self::new(name, llvm_usize)
+    pub fn new(name: String, size_t: u32) -> DefaultCodeGenerator {
+        assert!(matches!(size_t, 32 | 64));
+        DefaultCodeGenerator { name, size_t }
     }
 }
 
 impl CodeGenerator for DefaultCodeGenerator {
+    /// Returns the name for this [`CodeGenerator`].
     fn get_name(&self) -> &str {
         &self.name
     }
 
+    /// Returns an LLVM integer type representing `size_t`.
     fn get_size_type<'ctx>(&self, ctx: &'ctx Context) -> IntType<'ctx> {
         // it should be unsigned, but we don't really need unsigned and this could save us from
         // having to do a bit cast...

nac3core/src/codegen/irrt/error_context.rs

@@ -0,0 +1,187 @@
+use crate::codegen::{model::*, CodeGenContext, CodeGenerator};
+
+use super::util::{get_sized_dependent_function_name, FunctionBuilder};
+
+pub struct StrFields<'ctx> {
+    pub content: Field<PointerModel<FixedIntModel<Byte>>>,
+    pub length: Field<IntModel<'ctx>>,
+}
+
+#[derive(Debug, Clone, Copy)]
+pub struct Str<'ctx> {
+    pub sizet: IntModel<'ctx>,
+}
+
+impl<'ctx> IsStruct<'ctx> for Str<'ctx> {
+    type Fields = StrFields<'ctx>;
+
+    fn struct_name(&self) -> &'static str {
+        "Str"
+    }
+
+    fn build_fields(&self, builder: &mut FieldBuilder<'ctx>) -> Self::Fields {
+        Self::Fields {
+            content: builder.add_field_auto("content"),
+            length: builder.add_field("length", self.sizet),
+        }
+    }
+}
+
+type ErrorId = Int32;
+pub struct ErrorIdsFields {
+    pub index_error: Field<FixedIntModel<ErrorId>>,
+    pub value_error: Field<FixedIntModel<ErrorId>>,
+    pub assertion_error: Field<FixedIntModel<ErrorId>>,
+    pub runtime_error: Field<FixedIntModel<ErrorId>>,
+    pub type_error: Field<FixedIntModel<ErrorId>>,
+}
+
+#[derive(Debug, Clone, Copy)]
+pub struct ErrorIds;
+
+impl<'ctx> IsStruct<'ctx> for ErrorIds {
+    type Fields = ErrorIdsFields;
+
+    fn struct_name(&self) -> &'static str {
+        "ErrorIds"
+    }
+
+    fn build_fields(&self, builder: &mut FieldBuilder) -> Self::Fields {
+        Self::Fields {
+            index_error: builder.add_field_auto("index_error"),
+            value_error: builder.add_field_auto("value_error"),
+            assertion_error: builder.add_field_auto("assertion_error"),
+            runtime_error: builder.add_field_auto("runtime_error"),
+            type_error: builder.add_field_auto("type_error"),
+        }
+    }
+}
+
+pub struct ErrorContextFields {
+    pub error_id: Field<FixedIntModel<ErrorId>>,
+    pub message_template: Field<PointerModel<FixedIntModel<Byte>>>,
+    pub param1: Field<FixedIntModel<Int64>>,
+    pub param2: Field<FixedIntModel<Int64>>,
+    pub param3: Field<FixedIntModel<Int64>>,
+}
+
+#[derive(Debug, Clone, Copy)]
+pub struct ErrorContext;
+
+impl<'ctx> IsStruct<'ctx> for ErrorContext {
+    type Fields = ErrorContextFields;
+
+    fn struct_name(&self) -> &'static str {
+        "ErrorIds"
+    }
+
+    fn build_fields(&self, builder: &mut FieldBuilder) -> Self::Fields {
+        Self::Fields {
+            error_id: builder.add_field_auto("error_id"),
+            message_template: builder.add_field_auto("message_template"),
+            param1: builder.add_field_auto("param1"),
+            param2: builder.add_field_auto("param2"),
+            param3: builder.add_field_auto("param3"),
+        }
+    }
+}
+
+// Prepare ErrorIds
+fn build_error_ids<'ctx>(ctx: &CodeGenContext<'ctx, '_>) -> Pointer<'ctx, StructModel<ErrorIds>> {
+    // ErrorIdsLens.get_fields(ctx.ctx).assertion_error.
+    let error_ids = StructModel(ErrorIds).alloca(ctx, "error_ids");
+    let i32_model = FixedIntModel(Int32);
+    // i32_model.make_constant()
+
+    let get_string_id =
+        |string_id| i32_model.constant(ctx.ctx, ctx.resolver.get_string_id(string_id) as u64);
+
+    error_ids.gep(ctx, |f| f.index_error).store(ctx, get_string_id("0:IndexError"));
+    error_ids.gep(ctx, |f| f.value_error).store(ctx, get_string_id("0:ValueError"));
+    error_ids.gep(ctx, |f| f.assertion_error).store(ctx, get_string_id("0:AssertionError"));
+    error_ids.gep(ctx, |f| f.runtime_error).store(ctx, get_string_id("0:RuntimeError"));
+    error_ids.gep(ctx, |f| f.type_error).store(ctx, get_string_id("0:TypeError"));
+
+    error_ids
+}
+
+pub fn call_nac3_error_context_initialize<'ctx>(
+    ctx: &CodeGenContext<'ctx, '_>,
+    perrctx: Pointer<'ctx, StructModel<ErrorContext>>,
+    perror_ids: Pointer<'ctx, StructModel<ErrorIds>>,
+) {
+    FunctionBuilder::begin(ctx, "__nac3_error_context_initialize")
+        .arg("errctx", PointerModel(StructModel(ErrorContext)), perrctx)
+        .arg("error_ids", PointerModel(StructModel(ErrorIds)), perror_ids)
+        .returning_void();
+}
+
+pub fn call_nac3_error_context_has_no_error<'ctx>(
+    ctx: &CodeGenContext<'ctx, '_>,
+    errctx: Pointer<'ctx, StructModel<ErrorContext>>,
+) -> FixedInt<'ctx, Bool> {
+    FunctionBuilder::begin(ctx, "__nac3_error_context_has_no_error")
+        .arg("errctx", PointerModel(StructModel(ErrorContext)), errctx)
+        .returning("has_error", FixedIntModel(Bool))
+}
+
+pub fn call_nac3_error_context_get_error_str<'ctx>(
+    sizet: IntModel<'ctx>,
+    ctx: &CodeGenContext<'ctx, '_>,
+    errctx: Pointer<'ctx, StructModel<ErrorContext>>,
+    dst_str: Pointer<'ctx, StructModel<Str<'ctx>>>,
+) {
+    FunctionBuilder::begin(
+        ctx,
+        &get_sized_dependent_function_name(sizet, "__nac3_error_context_get_error_str"),
+    )
+    .arg("errctx", PointerModel(StructModel(ErrorContext)), errctx)
+    .arg("dst_str", PointerModel(StructModel(Str { sizet })), dst_str)
+    .returning_void();
+}
+
+pub fn prepare_error_context<'ctx>(
+    ctx: &CodeGenContext<'ctx, '_>,
+) -> Pointer<'ctx, StructModel<ErrorContext>> {
+    let error_ids = build_error_ids(ctx);
+    let errctx_ptr = StructModel(ErrorContext).alloca(ctx, "errctx");
+    call_nac3_error_context_initialize(ctx, errctx_ptr, error_ids);
+    errctx_ptr
+}
+
+pub fn check_error_context<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    errctx_ptr: Pointer<'ctx, StructModel<ErrorContext>>,
+) {
+    let sizet = IntModel(generator.get_size_type(ctx.ctx));
+
+    let has_error = call_nac3_error_context_has_no_error(ctx, errctx_ptr);
+    let pstr = StructModel(Str { sizet }).alloca(ctx, "error_str");
+    call_nac3_error_context_get_error_str(sizet, ctx, errctx_ptr, pstr);
+
+    let error_id = errctx_ptr.gep(ctx, |f| f.error_id).load(ctx, "error_id");
+    let error_str = pstr.load(ctx, "error_str");
+    let param1 = errctx_ptr.gep(ctx, |f| f.param1).load(ctx, "param1");
+    let param2 = errctx_ptr.gep(ctx, |f| f.param2).load(ctx, "param2");
+    let param3 = errctx_ptr.gep(ctx, |f| f.param3).load(ctx, "param3");
+    ctx.make_assert_impl_by_id(
+        generator,
+        has_error.value,
+        error_id.value,
+        error_str.get_llvm_value(),
+        [Some(param1.value), Some(param2.value), Some(param3.value)],
+        ctx.current_loc,
+    );
+}
+
+pub fn call_nac3_dummy_raise<G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext,
+) {
+    let errctx = prepare_error_context(ctx);
+    FunctionBuilder::begin(ctx, "__nac3_error_dummy_raise")
+        .arg("errctx", PointerModel(StructModel(ErrorContext)), errctx)
+        .returning_void();
+    check_error_context(generator, ctx, errctx);
+}

nac3core/src/codegen/irrt/list.rs

@@ -1,174 +0,0 @@
-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 llvm_usize = ctx.get_size_type();
-    let llvm_pi8 = ctx.ctx.i8_type().ptr_type(AddressSpace::default());
-    let llvm_i32 = ctx.ctx.i32_type();
-
-    assert_eq!(dest_idx.0.get_type(), llvm_i32);
-    assert_eq!(dest_idx.1.get_type(), llvm_i32);
-    assert_eq!(dest_idx.2.get_type(), llvm_i32);
-    assert_eq!(src_idx.0.get_type(), llvm_i32);
-    assert_eq!(src_idx.1.get_type(), llvm_i32);
-    assert_eq!(src_idx.2.get_type(), llvm_i32);
-
-    let (fun_symbol, elem_ptr_type) = ("__nac3_list_slice_assign_var_size", llvm_pi8);
-    let slice_assign_fun = {
-        let ty_vec = vec![
-            llvm_i32.into(),      // dest start idx
-            llvm_i32.into(),      // dest end idx
-            llvm_i32.into(),      // dest step
-            elem_ptr_type.into(), // dest arr ptr
-            llvm_i32.into(),      // dest arr len
-            llvm_i32.into(),      // src start idx
-            llvm_i32.into(),      // src end idx
-            llvm_i32.into(),      // src step
-            elem_ptr_type.into(), // src arr ptr
-            llvm_i32.into(),      // src arr len
-            llvm_i32.into(),      // size
-        ];
-        ctx.module.get_function(fun_symbol).unwrap_or_else(|| {
-            let fn_t = llvm_i32.fn_type(ty_vec.as_slice(), false);
-            ctx.module.add_function(fun_symbol, fn_t, None)
-        })
-    };
-
-    let zero = llvm_i32.const_zero();
-    let one = llvm_i32.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, llvm_i32, "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, llvm_i32, "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, llvm_i32, "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, llvm_usize, "new_len").unwrap();
-    dest_arr.store_size(ctx, new_len);
-    ctx.builder.build_unconditional_branch(cont_bb).unwrap();
-    ctx.builder.position_at_end(cont_bb);
-}

nac3core/src/codegen/irrt/math.rs

@@ -1,168 +0,0 @@
-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 llvm_i32 = ctx.ctx.i32_type();
-    let llvm_f64 = ctx.ctx.f64_type();
-
-    assert_eq!(v.get_type(), llvm_f64);
-
-    let intrinsic_fn = ctx.module.get_function("__nac3_isinf").unwrap_or_else(|| {
-        let fn_type = llvm_i32.fn_type(&[llvm_f64.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 llvm_i32 = ctx.ctx.i32_type();
-    let llvm_f64 = ctx.ctx.f64_type();
-
-    assert_eq!(v.get_type(), llvm_f64);
-
-    let intrinsic_fn = ctx.module.get_function("__nac3_isnan").unwrap_or_else(|| {
-        let fn_type = llvm_i32.fn_type(&[llvm_f64.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();
-
-    assert_eq!(v.get_type(), llvm_f64);
-
-    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();
-
-    assert_eq!(v.get_type(), llvm_f64);
-
-    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();
-
-    assert_eq!(v.get_type(), llvm_f64);
-
-    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

@@ -1,31 +1,33 @@
+use crate::typecheck::typedef::Type;
+
+pub mod error_context;
+pub mod numpy;
+mod test;
+mod util;
+
+use super::{
+    classes::{
+        ArrayLikeIndexer, ArrayLikeValue, ArraySliceValue, ListValue, NDArrayValue,
+        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,
-    values::{BasicValue, BasicValueEnum, IntValue},
-    IntPredicate,
+    types::{BasicTypeEnum, IntType},
+    values::{BasicValueEnum, CallSiteValue, FloatValue, IntValue},
+    AddressSpace, 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 range::*;
-pub use slice::*;
-pub use string::*;
-
-mod list;
-mod math;
-pub mod ndarray;
-mod range;
-mod slice;
-mod string;
-
 #[must_use]
-pub fn load_irrt<'ctx>(ctx: &'ctx Context, symbol_resolver: &dyn SymbolResolver) -> Module<'ctx> {
+pub fn load_irrt(ctx: &Context) -> Module {
     let bitcode_buf = MemoryBuffer::create_from_memory_range(
         include_bytes!(concat!(env!("OUT_DIR"), "/irrt.bc")),
         "irrt_bitcode_buffer",
@@ -41,43 +43,89 @@ pub fn load_irrt<'ctx>(ctx: &'ctx Context, symbol_resolver: &dyn SymbolResolver)
         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
 }
 
-/// Returns the name of a function which contains variants for 32-bit and 64-bit `size_t`.
-///
-/// - When [`TypeContext::size_type`] is 32-bits, the function name is `fn_name}`.
-/// - When [`TypeContext::size_type`] is 64-bits, the function name is `{fn_name}64`.
-#[must_use]
-pub fn get_usize_dependent_function_name(ctx: &CodeGenContext<'_, '_>, name: &str) -> String {
-    let mut name = name.to_owned();
-    match ctx.get_size_type().get_bit_width() {
-        32 => {}
-        64 => name.push_str("64"),
-        bit_width => {
-            panic!("Unsupported int type bit width {bit_width}, must be either 32-bits or 64-bits")
-        }
-    }
-    name
+// 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***,
@@ -128,11 +176,10 @@ pub fn handle_slice_indices<'ctx, G: CodeGenerator>(
     generator: &mut G,
     length: IntValue<'ctx>,
 ) -> Result<Option<(IntValue<'ctx>, IntValue<'ctx>, IntValue<'ctx>)>, String> {
-    let llvm_i32 = ctx.ctx.i32_type();
-
-    let zero = llvm_i32.const_zero();
-    let one = llvm_i32.const_int(1, false);
-    let length = ctx.builder.build_int_truncate_or_bit_cast(length, llvm_i32, "leni32").unwrap();
+    let int32 = ctx.ctx.i32_type();
+    let zero = int32.const_zero();
+    let one = int32.const_int(1, false);
+    let length = ctx.builder.build_int_truncate_or_bit_cast(length, int32, "leni32").unwrap();
     Ok(Some(match (start, end, step) {
         (s, e, None) => (
             if let Some(s) = s.as_ref() {
@@ -141,7 +188,7 @@ pub fn handle_slice_indices<'ctx, G: CodeGenerator>(
                     None => return Ok(None),
                 }
             } else {
-                llvm_i32.const_zero()
+                int32.const_zero()
             },
             {
                 let e = if let Some(s) = e.as_ref() {
@@ -246,3 +293,642 @@ 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()),
+    )
+}

nac3core/src/codegen/irrt/ndarray/array.rs

@@ -1,72 +0,0 @@
-use inkwell::{types::BasicTypeEnum, values::IntValue};
-
-use crate::codegen::{
-    expr::infer_and_call_function,
-    irrt::get_usize_dependent_function_name,
-    values::{ndarray::NDArrayValue, ListValue, ProxyValue, TypedArrayLikeAccessor},
-    CodeGenContext, CodeGenerator,
-};
-
-/// Generates a call to `__nac3_ndarray_array_set_and_validate_list_shape`.
-///
-/// Deduces the target shape of the `ndarray` from the provided `list`, raising an exception if
-/// there is any issue with the resultant `shape`.
-///
-/// `shape` must be pre-allocated by the caller of this function to `[usize; ndims]`, and must be
-/// initialized to all `-1`s.
-pub fn call_nac3_ndarray_array_set_and_validate_list_shape<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    list: ListValue<'ctx>,
-    ndims: IntValue<'ctx>,
-    shape: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-) {
-    let llvm_usize = ctx.get_size_type();
-    assert_eq!(list.get_type().element_type().unwrap(), ctx.ctx.i8_type().into());
-    assert_eq!(ndims.get_type(), llvm_usize);
-    assert_eq!(
-        BasicTypeEnum::try_from(shape.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-
-    let name =
-        get_usize_dependent_function_name(ctx, "__nac3_ndarray_array_set_and_validate_list_shape");
-
-    infer_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[list.as_base_value().into(), ndims.into(), shape.base_ptr(ctx, generator).into()],
-        None,
-        None,
-    );
-}
-
-/// Generates a call to `__nac3_ndarray_array_write_list_to_array`.
-///
-/// Copies the contents stored in `list` into `ndarray`.
-///
-/// The `ndarray` must fulfill the following preconditions:
-///
-/// - `ndarray.itemsize`: Must be initialized.
-/// - `ndarray.ndims`: Must be initialized.
-/// - `ndarray.shape`: Must be initialized.
-/// - `ndarray.data`: Must be allocated and contiguous.
-pub fn call_nac3_ndarray_array_write_list_to_array<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    list: ListValue<'ctx>,
-    ndarray: NDArrayValue<'ctx>,
-) {
-    assert_eq!(list.get_type().element_type().unwrap(), ctx.ctx.i8_type().into());
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_array_write_list_to_array");
-
-    infer_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[list.as_base_value().into(), ndarray.as_base_value().into()],
-        None,
-        None,
-    );
-}

nac3core/src/codegen/irrt/ndarray/basic.rs

@@ -1,295 +0,0 @@
-use inkwell::{
-    types::BasicTypeEnum,
-    values::{BasicValueEnum, IntValue, PointerValue},
-    AddressSpace,
-};
-
-use crate::codegen::{
-    expr::{create_and_call_function, infer_and_call_function},
-    irrt::get_usize_dependent_function_name,
-    types::ProxyType,
-    values::{ndarray::NDArrayValue, ProxyValue, TypedArrayLikeAccessor},
-    CodeGenContext, CodeGenerator,
-};
-
-/// Generates a call to `__nac3_ndarray_util_assert_shape_no_negative`.
-///
-/// Assets that `shape` does not contain negative dimensions.
-pub fn call_nac3_ndarray_util_assert_shape_no_negative<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    shape: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-) {
-    let llvm_usize = ctx.get_size_type();
-    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
-
-    assert_eq!(
-        BasicTypeEnum::try_from(shape.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-
-    let name =
-        get_usize_dependent_function_name(ctx, "__nac3_ndarray_util_assert_shape_no_negative");
-
-    create_and_call_function(
-        ctx,
-        &name,
-        Some(llvm_usize.into()),
-        &[
-            (llvm_usize.into(), shape.size(ctx, generator).into()),
-            (llvm_pusize.into(), shape.base_ptr(ctx, generator).into()),
-        ],
-        None,
-        None,
-    );
-}
-
-/// Generates a call to `__nac3_ndarray_util_assert_shape_output_shape_same`.
-///
-/// Asserts that `ndarray_shape` and `output_shape` are the same in the context of writing output to
-/// an `ndarray`.
-pub fn call_nac3_ndarray_util_assert_output_shape_same<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    ndarray_shape: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-    output_shape: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-) {
-    let llvm_usize = ctx.get_size_type();
-    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
-
-    assert_eq!(
-        BasicTypeEnum::try_from(ndarray_shape.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-    assert_eq!(
-        BasicTypeEnum::try_from(output_shape.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-
-    let name =
-        get_usize_dependent_function_name(ctx, "__nac3_ndarray_util_assert_output_shape_same");
-
-    create_and_call_function(
-        ctx,
-        &name,
-        Some(llvm_usize.into()),
-        &[
-            (llvm_usize.into(), ndarray_shape.size(ctx, generator).into()),
-            (llvm_pusize.into(), ndarray_shape.base_ptr(ctx, generator).into()),
-            (llvm_usize.into(), output_shape.size(ctx, generator).into()),
-            (llvm_pusize.into(), output_shape.base_ptr(ctx, generator).into()),
-        ],
-        None,
-        None,
-    );
-}
-
-/// Generates a call to `__nac3_ndarray_size`.
-///
-/// Returns a [`usize`][CodeGenerator::get_size_type] value of the number of elements of an
-/// `ndarray`, corresponding to the value of `ndarray.size`.
-pub fn call_nac3_ndarray_size<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    ndarray: NDArrayValue<'ctx>,
-) -> IntValue<'ctx> {
-    let llvm_usize = ctx.get_size_type();
-    let llvm_ndarray = ndarray.get_type().as_base_type();
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_size");
-
-    create_and_call_function(
-        ctx,
-        &name,
-        Some(llvm_usize.into()),
-        &[(llvm_ndarray.into(), ndarray.as_base_value().into())],
-        Some("size"),
-        None,
-    )
-    .map(BasicValueEnum::into_int_value)
-    .unwrap()
-}
-
-/// Generates a call to `__nac3_ndarray_nbytes`.
-///
-/// Returns a [`usize`][CodeGenerator::get_size_type] value of the number of bytes consumed by the
-/// data of the `ndarray`, corresponding to the value of `ndarray.nbytes`.
-pub fn call_nac3_ndarray_nbytes<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    ndarray: NDArrayValue<'ctx>,
-) -> IntValue<'ctx> {
-    let llvm_usize = ctx.get_size_type();
-    let llvm_ndarray = ndarray.get_type().as_base_type();
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_nbytes");
-
-    create_and_call_function(
-        ctx,
-        &name,
-        Some(llvm_usize.into()),
-        &[(llvm_ndarray.into(), ndarray.as_base_value().into())],
-        Some("nbytes"),
-        None,
-    )
-    .map(BasicValueEnum::into_int_value)
-    .unwrap()
-}
-
-/// Generates a call to `__nac3_ndarray_len`.
-///
-/// Returns a [`usize`][CodeGenerator::get_size_type] value of the size of the topmost dimension of
-/// the `ndarray`, corresponding to the value of `ndarray.__len__`.
-pub fn call_nac3_ndarray_len<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    ndarray: NDArrayValue<'ctx>,
-) -> IntValue<'ctx> {
-    let llvm_usize = ctx.get_size_type();
-    let llvm_ndarray = ndarray.get_type().as_base_type();
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_len");
-
-    create_and_call_function(
-        ctx,
-        &name,
-        Some(llvm_usize.into()),
-        &[(llvm_ndarray.into(), ndarray.as_base_value().into())],
-        Some("len"),
-        None,
-    )
-    .map(BasicValueEnum::into_int_value)
-    .unwrap()
-}
-
-/// Generates a call to `__nac3_ndarray_is_c_contiguous`.
-///
-/// Returns an `i1` value indicating whether the `ndarray` is C-contiguous.
-pub fn call_nac3_ndarray_is_c_contiguous<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    ndarray: NDArrayValue<'ctx>,
-) -> IntValue<'ctx> {
-    let llvm_i1 = ctx.ctx.bool_type();
-    let llvm_ndarray = ndarray.get_type().as_base_type();
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_is_c_contiguous");
-
-    create_and_call_function(
-        ctx,
-        &name,
-        Some(llvm_i1.into()),
-        &[(llvm_ndarray.into(), ndarray.as_base_value().into())],
-        Some("is_c_contiguous"),
-        None,
-    )
-    .map(BasicValueEnum::into_int_value)
-    .unwrap()
-}
-
-/// Generates a call to `__nac3_ndarray_get_nth_pelement`.
-///
-/// Returns a [`PointerValue`] to the `index`-th flattened element of the `ndarray`.
-pub fn call_nac3_ndarray_get_nth_pelement<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    ndarray: NDArrayValue<'ctx>,
-    index: IntValue<'ctx>,
-) -> PointerValue<'ctx> {
-    let llvm_i8 = ctx.ctx.i8_type();
-    let llvm_pi8 = llvm_i8.ptr_type(AddressSpace::default());
-    let llvm_usize = ctx.get_size_type();
-    let llvm_ndarray = ndarray.get_type().as_base_type();
-
-    assert_eq!(index.get_type(), llvm_usize);
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_get_nth_pelement");
-
-    create_and_call_function(
-        ctx,
-        &name,
-        Some(llvm_pi8.into()),
-        &[(llvm_ndarray.into(), ndarray.as_base_value().into()), (llvm_usize.into(), index.into())],
-        Some("pelement"),
-        None,
-    )
-    .map(BasicValueEnum::into_pointer_value)
-    .unwrap()
-}
-
-/// Generates a call to `__nac3_ndarray_get_pelement_by_indices`.
-///
-/// `indices` must have the same number of elements as the number of dimensions in `ndarray`.
-///
-/// Returns a [`PointerValue`] to the element indexed by `indices`.
-pub fn call_nac3_ndarray_get_pelement_by_indices<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    ndarray: NDArrayValue<'ctx>,
-    indices: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-) -> PointerValue<'ctx> {
-    let llvm_i8 = ctx.ctx.i8_type();
-    let llvm_pi8 = llvm_i8.ptr_type(AddressSpace::default());
-    let llvm_usize = ctx.get_size_type();
-    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
-    let llvm_ndarray = ndarray.get_type().as_base_type();
-
-    assert_eq!(
-        BasicTypeEnum::try_from(indices.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_get_pelement_by_indices");
-
-    create_and_call_function(
-        ctx,
-        &name,
-        Some(llvm_pi8.into()),
-        &[
-            (llvm_ndarray.into(), ndarray.as_base_value().into()),
-            (llvm_pusize.into(), indices.base_ptr(ctx, generator).into()),
-        ],
-        Some("pelement"),
-        None,
-    )
-    .map(BasicValueEnum::into_pointer_value)
-    .unwrap()
-}
-
-/// Generates a call to `__nac3_ndarray_set_strides_by_shape`.
-///
-/// Sets `ndarray.strides` assuming that `ndarray.shape` is C-contiguous.
-pub fn call_nac3_ndarray_set_strides_by_shape<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    ndarray: NDArrayValue<'ctx>,
-) {
-    let llvm_ndarray = ndarray.get_type().as_base_type();
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_set_strides_by_shape");
-
-    create_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[(llvm_ndarray.into(), ndarray.as_base_value().into())],
-        None,
-        None,
-    );
-}
-
-/// Generates a call to `__nac3_ndarray_copy_data`.
-///
-/// Copies all elements from `src_ndarray` to `dst_ndarray` using their flattened views. The number
-/// of elements in `src_ndarray` must be greater than or equal to the number of elements in
-/// `dst_ndarray`.
-pub fn call_nac3_ndarray_copy_data<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    src_ndarray: NDArrayValue<'ctx>,
-    dst_ndarray: NDArrayValue<'ctx>,
-) {
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_copy_data");
-
-    infer_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[src_ndarray.as_base_value().into(), dst_ndarray.as_base_value().into()],
-        None,
-        None,
-    );
-}

nac3core/src/codegen/irrt/ndarray/broadcast.rs

@@ -1,81 +0,0 @@
-use inkwell::values::IntValue;
-
-use crate::codegen::{
-    expr::infer_and_call_function,
-    irrt::get_usize_dependent_function_name,
-    types::{ndarray::ShapeEntryType, ProxyType},
-    values::{
-        ndarray::NDArrayValue, ArrayLikeValue, ArraySliceValue, ProxyValue, TypedArrayLikeAccessor,
-        TypedArrayLikeMutator,
-    },
-    CodeGenContext, CodeGenerator,
-};
-
-/// Generates a call to `__nac3_ndarray_broadcast_to`.
-///
-/// Attempts to broadcast `src_ndarray` to the new shape defined by `dst_ndarray`.
-///
-/// `dst_ndarray` must meet the following preconditions:
-///
-/// - `dst_ndarray.ndims` must be initialized and matching the length of `dst_ndarray.shape`.
-/// - `dst_ndarray.shape` must be initialized and contains the target broadcast shape.
-/// - `dst_ndarray.strides` must be allocated and may contain uninitialized values.
-pub fn call_nac3_ndarray_broadcast_to<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    src_ndarray: NDArrayValue<'ctx>,
-    dst_ndarray: NDArrayValue<'ctx>,
-) {
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_broadcast_to");
-    infer_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[src_ndarray.as_base_value().into(), dst_ndarray.as_base_value().into()],
-        None,
-        None,
-    );
-}
-
-/// Generates a call to `__nac3_ndarray_broadcast_shapes`.
-///
-/// Attempts to calculate the resultant shape from broadcasting all shapes in `shape_entries`,
-/// writing the result to `dst_shape`.
-pub fn call_nac3_ndarray_broadcast_shapes<'ctx, G, Shape>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    num_shape_entries: IntValue<'ctx>,
-    shape_entries: ArraySliceValue<'ctx>,
-    dst_ndims: IntValue<'ctx>,
-    dst_shape: &Shape,
-) where
-    G: CodeGenerator + ?Sized,
-    Shape: TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>
-        + TypedArrayLikeMutator<'ctx, G, IntValue<'ctx>>,
-{
-    let llvm_usize = ctx.get_size_type();
-
-    assert_eq!(num_shape_entries.get_type(), llvm_usize);
-    assert!(ShapeEntryType::is_type(
-        generator,
-        ctx.ctx,
-        shape_entries.base_ptr(ctx, generator).get_type()
-    )
-    .is_ok());
-    assert_eq!(dst_ndims.get_type(), llvm_usize);
-    assert_eq!(dst_shape.element_type(ctx, generator), llvm_usize.into());
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_broadcast_shapes");
-    infer_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[
-            num_shape_entries.into(),
-            shape_entries.base_ptr(ctx, generator).into(),
-            dst_ndims.into(),
-            dst_shape.base_ptr(ctx, generator).into(),
-        ],
-        None,
-        None,
-    );
-}

nac3core/src/codegen/irrt/ndarray/indexing.rs

@@ -1,34 +0,0 @@
-use crate::codegen::{
-    expr::infer_and_call_function,
-    irrt::get_usize_dependent_function_name,
-    values::{ndarray::NDArrayValue, ArrayLikeValue, ArraySliceValue, ProxyValue},
-    CodeGenContext, CodeGenerator,
-};
-
-/// Generates a call to `__nac3_ndarray_index`.
-///
-/// Performs [basic indexing](https://numpy.org/doc/stable/user/basics.indexing.html#basic-indexing)
-/// on `src_ndarray` using `indices`, writing the result to `dst_ndarray`, corresponding to the
-/// operation `dst_ndarray = src_ndarray[indices]`.
-pub fn call_nac3_ndarray_index<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    indices: ArraySliceValue<'ctx>,
-    src_ndarray: NDArrayValue<'ctx>,
-    dst_ndarray: NDArrayValue<'ctx>,
-) {
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_index");
-    infer_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[
-            indices.size(ctx, generator).into(),
-            indices.base_ptr(ctx, generator).into(),
-            src_ndarray.as_base_value().into(),
-            dst_ndarray.as_base_value().into(),
-        ],
-        None,
-        None,
-    );
-}

nac3core/src/codegen/irrt/ndarray/iter.rs

@@ -1,81 +0,0 @@
-use inkwell::{
-    types::BasicTypeEnum,
-    values::{BasicValueEnum, IntValue},
-    AddressSpace,
-};
-
-use crate::codegen::{
-    expr::{create_and_call_function, infer_and_call_function},
-    irrt::get_usize_dependent_function_name,
-    types::ProxyType,
-    values::{
-        ndarray::{NDArrayValue, NDIterValue},
-        ProxyValue, TypedArrayLikeAccessor,
-    },
-    CodeGenContext, CodeGenerator,
-};
-
-/// Generates a call to `__nac3_nditer_initialize`.
-///
-/// Initializes the `iter` object.
-pub fn call_nac3_nditer_initialize<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    iter: NDIterValue<'ctx>,
-    ndarray: NDArrayValue<'ctx>,
-    indices: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-) {
-    let llvm_usize = ctx.get_size_type();
-    let llvm_pusize = llvm_usize.ptr_type(AddressSpace::default());
-
-    assert_eq!(
-        BasicTypeEnum::try_from(indices.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_nditer_initialize");
-
-    create_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[
-            (iter.get_type().as_base_type().into(), iter.as_base_value().into()),
-            (ndarray.get_type().as_base_type().into(), ndarray.as_base_value().into()),
-            (llvm_pusize.into(), indices.base_ptr(ctx, generator).into()),
-        ],
-        None,
-        None,
-    );
-}
-
-/// Generates a call to `__nac3_nditer_initialize_has_element`.
-///
-/// Returns an `i1` value indicating whether there are elements left to traverse for the `iter`
-/// object.
-pub fn call_nac3_nditer_has_element<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    iter: NDIterValue<'ctx>,
-) -> IntValue<'ctx> {
-    let name = get_usize_dependent_function_name(ctx, "__nac3_nditer_has_element");
-
-    infer_and_call_function(
-        ctx,
-        &name,
-        Some(ctx.ctx.bool_type().into()),
-        &[iter.as_base_value().into()],
-        None,
-        None,
-    )
-    .map(BasicValueEnum::into_int_value)
-    .unwrap()
-}
-
-/// Generates a call to `__nac3_nditer_next`.
-///
-/// Moves `iter` to point to the next element.
-pub fn call_nac3_nditer_next<'ctx>(ctx: &CodeGenContext<'ctx, '_>, iter: NDIterValue<'ctx>) {
-    let name = get_usize_dependent_function_name(ctx, "__nac3_nditer_next");
-
-    infer_and_call_function(ctx, &name, None, &[iter.as_base_value().into()], None, None);
-}

nac3core/src/codegen/irrt/ndarray/matmul.rs

@@ -1,65 +0,0 @@
-use inkwell::{types::BasicTypeEnum, values::IntValue};
-
-use crate::codegen::{
-    expr::infer_and_call_function, irrt::get_usize_dependent_function_name,
-    values::TypedArrayLikeAccessor, CodeGenContext, CodeGenerator,
-};
-
-/// Generates a call to `__nac3_ndarray_matmul_calculate_shapes`.
-///
-/// Calculates the broadcasted shapes for `a`, `b`, and the `ndarray` holding the final values of
-/// `a @ b`.
-#[allow(clippy::too_many_arguments)]
-pub fn call_nac3_ndarray_matmul_calculate_shapes<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    a_shape: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-    b_shape: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-    final_ndims: IntValue<'ctx>,
-    new_a_shape: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-    new_b_shape: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-    dst_shape: &impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>,
-) {
-    let llvm_usize = ctx.get_size_type();
-
-    assert_eq!(
-        BasicTypeEnum::try_from(a_shape.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-    assert_eq!(
-        BasicTypeEnum::try_from(b_shape.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-    assert_eq!(
-        BasicTypeEnum::try_from(new_a_shape.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-    assert_eq!(
-        BasicTypeEnum::try_from(new_b_shape.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-    assert_eq!(
-        BasicTypeEnum::try_from(dst_shape.element_type(ctx, generator)).unwrap(),
-        llvm_usize.into()
-    );
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_matmul_calculate_shapes");
-
-    infer_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[
-            a_shape.size(ctx, generator).into(),
-            a_shape.base_ptr(ctx, generator).into(),
-            b_shape.size(ctx, generator).into(),
-            b_shape.base_ptr(ctx, generator).into(),
-            final_ndims.into(),
-            new_a_shape.base_ptr(ctx, generator).into(),
-            new_b_shape.base_ptr(ctx, generator).into(),
-            dst_shape.base_ptr(ctx, generator).into(),
-        ],
-        None,
-        None,
-    );
-}

nac3core/src/codegen/irrt/ndarray/mod.rs

@@ -1,17 +0,0 @@
-pub use array::*;
-pub use basic::*;
-pub use broadcast::*;
-pub use indexing::*;
-pub use iter::*;
-pub use matmul::*;
-pub use reshape::*;
-pub use transpose::*;
-
-mod array;
-mod basic;
-mod broadcast;
-mod indexing;
-mod iter;
-mod matmul;
-mod reshape;
-mod transpose;

nac3core/src/codegen/irrt/ndarray/reshape.rs

@@ -1,39 +0,0 @@
-use inkwell::values::IntValue;
-
-use crate::codegen::{
-    expr::infer_and_call_function,
-    irrt::get_usize_dependent_function_name,
-    values::{ArrayLikeValue, ArraySliceValue},
-    CodeGenContext, CodeGenerator,
-};
-
-/// Generates a call to `__nac3_ndarray_reshape_resolve_and_check_new_shape`.
-///
-/// Resolves unknown dimensions in `new_shape` for `numpy.reshape(<ndarray>, new_shape)`, raising an
-/// assertion if multiple dimensions are unknown (`-1`).
-pub fn call_nac3_ndarray_reshape_resolve_and_check_new_shape<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    size: IntValue<'ctx>,
-    new_ndims: IntValue<'ctx>,
-    new_shape: ArraySliceValue<'ctx>,
-) {
-    let llvm_usize = ctx.get_size_type();
-
-    assert_eq!(size.get_type(), llvm_usize);
-    assert_eq!(new_ndims.get_type(), llvm_usize);
-    assert_eq!(new_shape.element_type(ctx, generator), llvm_usize.into());
-
-    let name = get_usize_dependent_function_name(
-        ctx,
-        "__nac3_ndarray_reshape_resolve_and_check_new_shape",
-    );
-    infer_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[size.into(), new_ndims.into(), new_shape.base_ptr(ctx, generator).into()],
-        None,
-        None,
-    );
-}

nac3core/src/codegen/irrt/ndarray/transpose.rs

@@ -1,48 +0,0 @@
-use inkwell::{values::IntValue, AddressSpace};
-
-use crate::codegen::{
-    expr::infer_and_call_function,
-    irrt::get_usize_dependent_function_name,
-    values::{ndarray::NDArrayValue, ProxyValue, TypedArrayLikeAccessor},
-    CodeGenContext, CodeGenerator,
-};
-
-/// Generates a call to `__nac3_ndarray_transpose`.
-///
-/// Creates a transpose view of `src_ndarray` and writes the result to `dst_ndarray`.
-///
-/// `dst_ndarray` must fulfill the following preconditions:
-///
-/// - `dst_ndarray.ndims` must be initialized and must be equal to `src_ndarray.ndims`.
-/// - `dst_ndarray.shape` must be allocated and may contain uninitialized values.
-/// - `dst_ndarray.strides` must be allocated and may contain uninitialized values.
-pub fn call_nac3_ndarray_transpose<'ctx, G: CodeGenerator + ?Sized>(
-    generator: &G,
-    ctx: &CodeGenContext<'ctx, '_>,
-    src_ndarray: NDArrayValue<'ctx>,
-    dst_ndarray: NDArrayValue<'ctx>,
-    axes: Option<&impl TypedArrayLikeAccessor<'ctx, G, IntValue<'ctx>>>,
-) {
-    let llvm_usize = ctx.get_size_type();
-
-    assert!(axes.is_none_or(|axes| axes.size(ctx, generator).get_type() == llvm_usize));
-    assert!(axes.is_none_or(|axes| axes.element_type(ctx, generator) == llvm_usize.into()));
-
-    let name = get_usize_dependent_function_name(ctx, "__nac3_ndarray_transpose");
-    infer_and_call_function(
-        ctx,
-        &name,
-        None,
-        &[
-            src_ndarray.as_base_value().into(),
-            dst_ndarray.as_base_value().into(),
-            axes.map_or(llvm_usize.const_zero(), |axes| axes.size(ctx, generator)).into(),
-            axes.map_or(llvm_usize.ptr_type(AddressSpace::default()).const_null(), |axes| {
-                axes.base_ptr(ctx, generator)
-            })
-            .into(),
-        ],
-        None,
-        None,
-    );
-}

nac3core/src/codegen/irrt/numpy/mod.rs

@@ -0,0 +1,4 @@
+pub mod ndarray;
+pub mod shape;
+pub mod slice;
+pub mod subscript;

nac3core/src/codegen/irrt/numpy/ndarray.rs

@@ -0,0 +1,254 @@
+use inkwell::types::{BasicType, BasicTypeEnum};
+
+use crate::codegen::{
+    irrt::{
+        error_context::{check_error_context, prepare_error_context, ErrorContext},
+        util::{get_sized_dependent_function_name, FunctionBuilder},
+    },
+    model::*,
+    CodeGenContext, CodeGenerator,
+};
+
+use super::{
+    shape::Producer,
+    slice::{SliceIndex, SliceIndexModel},
+};
+
+pub struct NpArrayFields<'ctx> {
+    pub data: Field<PointerModel<ByteModel>>,
+    pub itemsize: Field<IntModel<'ctx>>,
+    pub ndims: Field<IntModel<'ctx>>,
+    pub shape: Field<PointerModel<IntModel<'ctx>>>,
+    pub strides: Field<PointerModel<IntModel<'ctx>>>,
+}
+
+#[derive(Debug, Clone, Copy)]
+pub struct NpArray<'ctx> {
+    pub sizet: IntModel<'ctx>,
+}
+
+impl<'ctx> IsStruct<'ctx> for NpArray<'ctx> {
+    type Fields = NpArrayFields<'ctx>;
+
+    fn struct_name(&self) -> &'static str {
+        "NDArray"
+    }
+
+    fn build_fields(&self, builder: &mut FieldBuilder<'ctx>) -> Self::Fields {
+        NpArrayFields {
+            data: builder.add_field_auto("data"),
+            itemsize: builder.add_field("itemsize", self.sizet),
+            ndims: builder.add_field("ndims", self.sizet),
+            shape: builder.add_field("shape", PointerModel(self.sizet)),
+            strides: builder.add_field("strides", PointerModel(self.sizet)),
+        }
+    }
+}
+
+impl<'ctx> Pointer<'ctx, StructModel<NpArray<'ctx>>> {
+    pub fn shape_slice(&self, ctx: &CodeGenContext<'ctx, '_>) -> ArraySlice<'ctx, IntModel<'ctx>> {
+        let ndims = self.gep(ctx, |f| f.ndims).load(ctx, "ndims");
+        let shape_base_ptr = self.gep(ctx, |f| f.shape).load(ctx, "shape");
+        ArraySlice { num_elements: ndims, pointer: shape_base_ptr }
+    }
+
+    pub fn strides_slice(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+    ) -> ArraySlice<'ctx, IntModel<'ctx>> {
+        let ndims = self.gep(ctx, |f| f.ndims).load(ctx, "ndims");
+        let strides_base_ptr = self.gep(ctx, |f| f.strides).load(ctx, "strides");
+        ArraySlice { num_elements: ndims, pointer: strides_base_ptr }
+    }
+}
+
+pub fn alloca_ndarray<'ctx, G>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    elem_type: BasicTypeEnum<'ctx>,
+    ndims: Int<'ctx>,
+    name: &str,
+) -> Result<Pointer<'ctx, StructModel<NpArray<'ctx>>>, String>
+where
+    G: CodeGenerator + ?Sized,
+{
+    let sizet = IntModel(generator.get_size_type(ctx.ctx));
+
+    // Allocate ndarray
+    let ndarray_ptr = StructModel(NpArray { sizet }).alloca(ctx, name);
+
+    // Set ndims
+    ndarray_ptr.gep(ctx, |f| f.ndims).store(ctx, ndims);
+
+    // Set itemsize
+    let itemsize = Int(elem_type.size_of().unwrap());
+    ndarray_ptr.gep(ctx, |f| f.itemsize).store(ctx, itemsize.signed_cast_to_int(ctx, sizet, ""));
+
+    // Allocate and set shape
+    let shape_array = sizet.array_alloca(ctx, ndims, "shape");
+    ndarray_ptr.gep(ctx, |f| f.shape).store(ctx, shape_array.pointer);
+
+    // Allocate and set strides
+    let strides_array = sizet.array_alloca(ctx, ndims, "strides");
+    ndarray_ptr.gep(ctx, |f| f.strides).store(ctx, strides_array.pointer);
+
+    Ok(ndarray_ptr)
+}
+
+pub enum NDArrayInitMode<'ctx, G: CodeGenerator + ?Sized> {
+    NDims { ndims: Int<'ctx> },
+    Shape { shape: Producer<'ctx, G, IntModel<'ctx>> },
+    ShapeAndAllocaData { shape: Producer<'ctx, G, IntModel<'ctx>> },
+}
+
+/// TODO: DOCUMENT ME
+pub fn alloca_ndarray_and_init<'ctx, G>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    elem_type: BasicTypeEnum<'ctx>,
+    init_mode: NDArrayInitMode<'ctx, G>,
+    name: &str,
+) -> Result<Pointer<'ctx, StructModel<NpArray<'ctx>>>, String>
+where
+    G: CodeGenerator + ?Sized,
+{
+    // It is implemented verbosely in order to make the initialization modes super clear in their intent.
+    match init_mode {
+        NDArrayInitMode::NDims { ndims } => {
+            let ndarray_ptr = alloca_ndarray(generator, ctx, elem_type, ndims, name)?;
+            Ok(ndarray_ptr)
+        }
+        NDArrayInitMode::Shape { shape } => {
+            let ndims = shape.count;
+            let ndarray_ptr = alloca_ndarray(generator, ctx, elem_type, ndims, name)?;
+
+            // Fill `ndarray.shape`
+            (shape.write_to_array)(generator, ctx, &ndarray_ptr.shape_slice(ctx))?;
+
+            // Check if `shape` has bad inputs
+            call_nac3_ndarray_util_assert_shape_no_negative(
+                generator,
+                ctx,
+                ndims,
+                ndarray_ptr.gep(ctx, |f| f.shape).load(ctx, "shape"),
+            );
+
+            // NOTE: DO NOT DO `set_strides_by_shape` HERE.
+            // Simply this is because we specified that `SetShape` wouldn't do `set_strides_by_shape`
+
+            Ok(ndarray_ptr)
+        }
+        NDArrayInitMode::ShapeAndAllocaData { shape } => {
+            let ndims = shape.count;
+            let ndarray_ptr = alloca_ndarray(generator, ctx, elem_type, ndims, name)?;
+
+            // Fill `ndarray.shape`
+            (shape.write_to_array)(generator, ctx, &ndarray_ptr.shape_slice(ctx))?;
+
+            // Check if `shape` has bad inputs
+            call_nac3_ndarray_util_assert_shape_no_negative(
+                generator,
+                ctx,
+                ndims,
+                ndarray_ptr.gep(ctx, |f| f.shape).load(ctx, "shape"),
+            );
+
+            // Now we populate `ndarray.data` by alloca-ing.
+            // But first, we need to know the size of the ndarray to know how many elements to alloca,
+            // since calculating nbytes of an ndarray requires `ndarray.shape` to be set.
+            let ndarray_nbytes = call_nac3_ndarray_nbytes(ctx, ndarray_ptr);
+
+            // Alloca `data` and assign it to `ndarray.data`
+            let data_array = FixedIntModel(Byte).array_alloca(ctx, ndarray_nbytes, "data");
+            ndarray_ptr.gep(ctx, |f| f.data).store(ctx, data_array.pointer);
+
+            // Finally, do `set_strides_by_shape`
+            // Check out https://ajcr.net/stride-guide-part-1/ to see what numpy "strides" are.
+            call_nac3_ndarray_set_strides_by_shape(ctx, ndarray_ptr);
+
+            Ok(ndarray_ptr)
+        }
+    }
+}
+
+fn call_nac3_ndarray_util_assert_shape_no_negative<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    ndims: Int<'ctx>,
+    shape_ptr: Pointer<'ctx, IntModel<'ctx>>,
+) {
+    let sizet = IntModel(generator.get_size_type(ctx.ctx));
+
+    let errctx = prepare_error_context(ctx);
+    FunctionBuilder::begin(
+        ctx,
+        &get_sized_dependent_function_name(sizet, "__nac3_ndarray_util_assert_shape_no_negative"),
+    )
+    .arg("errctx", PointerModel(StructModel(ErrorContext)), errctx)
+    .arg("ndims", sizet, ndims)
+    .arg("shape", PointerModel(sizet), shape_ptr)
+    .returning_void();
+    check_error_context(generator, ctx, errctx);
+}
+
+fn call_nac3_ndarray_set_strides_by_shape<'ctx>(
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    ndarray_ptr: Pointer<'ctx, StructModel<NpArray<'ctx>>>,
+) {
+    let sizet = ndarray_ptr.element.0.sizet;
+
+    FunctionBuilder::begin(
+        ctx,
+        &get_sized_dependent_function_name(sizet, "__nac3_ndarray_set_strides_by_shape"),
+    )
+    .arg("ndarray", PointerModel(StructModel(NpArray { sizet })), ndarray_ptr)
+    .returning_void();
+}
+
+pub fn call_nac3_ndarray_nbytes<'ctx>(
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    ndarray_ptr: Pointer<'ctx, StructModel<NpArray<'ctx>>>,
+) -> Int<'ctx> {
+    let sizet = ndarray_ptr.element.0.sizet;
+
+    FunctionBuilder::begin(ctx, &get_sized_dependent_function_name(sizet, "__nac3_ndarray_nbytes"))
+        .arg("ndarray", PointerModel(StructModel(NpArray { sizet })), ndarray_ptr)
+        .returning("nbytes", sizet)
+}
+
+pub fn call_nac3_ndarray_fill_generic<'ctx>(
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    ndarray_ptr: Pointer<'ctx, StructModel<NpArray<'ctx>>>,
+    fill_value_ptr: Pointer<'ctx, ByteModel>,
+) {
+    let sizet = ndarray_ptr.element.0.sizet;
+
+    FunctionBuilder::begin(
+        ctx,
+        &get_sized_dependent_function_name(sizet, "__nac3_ndarray_fill_generic"),
+    )
+    .arg("ndarray", PointerModel(StructModel(NpArray { sizet })), ndarray_ptr)
+    .arg("pvalue", PointerModel(FixedIntModel(Byte)), fill_value_ptr)
+    .returning_void();
+}
+
+pub fn call_nac3_ndarray_len<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    ndarray_ptr: Pointer<'ctx, StructModel<NpArray<'ctx>>>,
+) -> SliceIndex<'ctx> {
+    let sizet = ndarray_ptr.element.0.sizet;
+    let slice_index_model = SliceIndexModel::default();
+
+    let dst_len = slice_index_model.alloca(ctx, "dst_len");
+
+    let errctx = prepare_error_context(ctx);
+    FunctionBuilder::begin(ctx, &get_sized_dependent_function_name(sizet, "__nac3_ndarray_len"))
+        .arg("errctx", PointerModel(StructModel(ErrorContext)), errctx)
+        .arg("ndarray", PointerModel(StructModel(NpArray { sizet })), ndarray_ptr)
+        .arg("dst_len", PointerModel(slice_index_model), dst_len)
+        .returning_void();
+    check_error_context(generator, ctx, errctx);
+
+    dst_len.load(ctx, "len")
+}

nac3core/src/codegen/irrt/numpy/shape.rs

@@ -0,0 +1,162 @@
+use inkwell::values::BasicValueEnum;
+
+use crate::{
+    codegen::{
+        classes::{ListValue, UntypedArrayLikeAccessor},
+        model::*,
+        stmt::gen_for_callback_incrementing,
+        CodeGenContext, CodeGenerator,
+    },
+    typecheck::typedef::{Type, TypeEnum},
+};
+
+pub type ProducerWriteToArray<'ctx, G, E> = Box<
+    dyn Fn(&mut G, &mut CodeGenContext<'ctx, '_>, &ArraySlice<'ctx, E>) -> Result<(), String>
+        + 'ctx,
+>;
+
+pub struct Producer<'ctx, G: CodeGenerator + ?Sized, E: Model<'ctx>> {
+    pub count: Int<'ctx>,
+    pub write_to_array: ProducerWriteToArray<'ctx, G, E>,
+}
+
+/// TODO: UPDATE DOCUMENTATION
+/// LLVM-typed implementation for generating a [`Producer`] that sets a list of ints.
+///
+/// * `elem_ty` - The element type of the `NDArray`.
+/// * `shape` - The `shape` parameter used to construct the `NDArray`.
+///
+/// ### Notes on `shape`
+///
+/// Just like numpy, the `shape` argument can be:
+///   1. A list of `int32`;   e.g., `np.empty([600, 800, 3])`
+///   2. A tuple of `int32`;  e.g., `np.empty((600, 800, 3))`
+///   3. A scalar `int32`;     e.g., `np.empty(3)`, this is functionally equivalent to `np.empty([3])`
+///
+/// See also [`typecheck::type_inferencer::fold_numpy_function_call_shape_argument`] to
+/// learn how `shape` gets from being a Python user expression to here.
+pub fn parse_input_shape_arg<'ctx, G>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    shape: BasicValueEnum<'ctx>,
+    shape_ty: Type,
+) -> Producer<'ctx, G, IntModel<'ctx>>
+where
+    G: CodeGenerator + ?Sized,
+{
+    let sizet = IntModel(generator.get_size_type(ctx.ctx));
+
+    match &*ctx.unifier.get_ty(shape_ty) {
+        TypeEnum::TObj { obj_id, .. }
+            if *obj_id == ctx.primitives.list.obj_id(&ctx.unifier).unwrap() =>
+        {
+            // 1. A list of ints; e.g., `np.empty([600, 800, 3])`
+
+            // A list has to be a PointerValue
+            let shape_list = ListValue::from_ptr_val(shape.into_pointer_value(), sizet.0, None);
+
+            // Create `Producer`
+            let ndims = Int(shape_list.load_size(ctx, Some("count")));
+            Producer {
+                count: ndims,
+                write_to_array: Box::new(move |ctx, generator, dst_array| {
+                    // Basically iterate through the list and write to `dst_slice` accordingly
+                    let init_val = sizet.constant(0).0;
+                    let max_val = (ndims.0, false);
+                    let incr_val = sizet.constant(1).0;
+                    gen_for_callback_incrementing(
+                        ctx,
+                        generator,
+                        init_val,
+                        max_val,
+                        |generator, ctx, _hooks, axis| {
+                            let axis = Int(axis);
+
+                            // Get the dimension at `axis`
+                            let dim = shape_list
+                                .data()
+                                .get(ctx, generator, &axis.0, None)
+                                .into_int_value();
+
+                            // Cast `dim` to SizeT
+                            let dim = ctx
+                                .builder
+                                .build_int_s_extend_or_bit_cast(dim, sizet.0, "dim_casted")
+                                .unwrap();
+
+                            // Write
+                            dst_array.ix(generator, ctx, axis, "dim").store(ctx, Int(dim));
+                            Ok(())
+                        },
+                        incr_val,
+                    )
+                }),
+            }
+        }
+        TypeEnum::TTuple { ty: tuple_types } => {
+            // 2. A tuple of ints; e.g., `np.empty((600, 800, 3))`
+
+            // Get the length/size of the tuple, which also happens to be the value of `ndims`.
+            let ndims = tuple_types.len();
+
+            // A tuple has to be a StructValue
+            // Read [`codegen::expr::gen_expr`] to see how `nac3core` translates a Python tuple into LLVM.
+            let shape_tuple = shape.into_struct_value();
+
+            Producer {
+                count: sizet.constant(ndims as u64),
+                write_to_array: Box::new(move |generator, ctx, dst_array| {
+                    for axis in 0..ndims {
+                        // Get the dimension at `axis`
+                        let dim = ctx
+                            .builder
+                            .build_extract_value(
+                                shape_tuple,
+                                axis as u32,
+                                format!("dim{axis}").as_str(),
+                            )
+                            .unwrap()
+                            .into_int_value();
+
+                        // Cast `dim` to SizeT
+                        let dim = ctx
+                            .builder
+                            .build_int_s_extend_or_bit_cast(dim, sizet.0, "dim_casted")
+                            .unwrap();
+
+                        // Write
+                        dst_array
+                            .ix(generator, ctx, sizet.constant(axis as u64), "dim")
+                            .store(ctx, Int(dim));
+                    }
+                    Ok(())
+                }),
+            }
+        }
+        TypeEnum::TObj { obj_id, .. }
+            if *obj_id == ctx.primitives.int32.obj_id(&ctx.unifier).unwrap() =>
+        {
+            // 3. A scalar int; e.g., `np.empty(3)`, this is functionally equivalent to `np.empty([3])`
+
+            // The value has to be an integer
+            let shape_int = shape.into_int_value();
+
+            Producer {
+                count: sizet.constant(1),
+                write_to_array: Box::new(move |generator, ctx, dst_array| {
+                    // Cast `shape_int` to SizeT
+                    let dim = ctx
+                        .builder
+                        .build_int_s_extend_or_bit_cast(shape_int, sizet.0, "dim_casted")
+                        .unwrap();
+
+                    // Set shape[0] = shape_int
+                    dst_array.ix(generator, ctx, sizet.constant(0), "dim").store(ctx, Int(dim));
+
+                    Ok(())
+                }),
+            }
+        }
+        _ => panic!("parse_input_shape_arg encountered unknown type"),
+    }
+}

nac3core/src/codegen/irrt/numpy/slice.rs

@@ -0,0 +1,86 @@
+use crate::codegen::{model::*, CodeGenContext};
+
+// nac3core's slicing index/length values are always int32_t
+pub type SliceIndexInt = Int32;
+pub type SliceIndexModel = FixedIntModel<SliceIndexInt>;
+pub type SliceIndex<'ctx> = FixedInt<'ctx, SliceIndexInt>;
+
+#[derive(Debug, Clone)]
+pub struct UserSliceFields {
+    pub start_defined: Field<BoolModel>,
+    pub start: Field<SliceIndexModel>,
+    pub stop_defined: Field<BoolModel>,
+    pub stop: Field<SliceIndexModel>,
+    pub step_defined: Field<BoolModel>,
+    pub step: Field<SliceIndexModel>,
+}
+
+#[derive(Debug, Clone, Copy, Default)]
+pub struct UserSlice;
+
+impl<'ctx> IsStruct<'ctx> for UserSlice {
+    type Fields = UserSliceFields;
+
+    fn struct_name(&self) -> &'static str {
+        "UserSlice"
+    }
+
+    fn build_fields(&self, builder: &mut FieldBuilder<'ctx>) -> Self::Fields {
+        Self::Fields {
+            start_defined: builder.add_field_auto("start_defined"),
+            start: builder.add_field_auto("start"),
+            stop_defined: builder.add_field_auto("stop_defined"),
+            stop: builder.add_field_auto("stop"),
+            step_defined: builder.add_field_auto("step_defined"),
+            step: builder.add_field_auto("step"),
+        }
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct RustUserSlice<'ctx> {
+    pub start: Option<SliceIndex<'ctx>>,
+    pub stop: Option<SliceIndex<'ctx>>,
+    pub step: Option<SliceIndex<'ctx>>,
+}
+
+impl<'ctx> RustUserSlice<'ctx> {
+    // Set the values of an LLVM UserSlice
+    // in the format of Python's `slice()`
+    pub fn write_to_user_slice(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        dst_slice_ptr: Pointer<'ctx, StructModel<UserSlice>>,
+    ) {
+        // TODO: make this neater, with a helper lambda?
+
+        let bool_model = BoolModel::default();
+
+        let false_ = bool_model.constant(ctx.ctx, 0);
+        let true_ = bool_model.constant(ctx.ctx, 1);
+
+        match self.start {
+            Some(start) => {
+                dst_slice_ptr.gep(ctx, |f| f.start_defined).store(ctx, true_);
+                dst_slice_ptr.gep(ctx, |f| f.start).store(ctx, start);
+            }
+            None => dst_slice_ptr.gep(ctx, |f| f.start_defined).store(ctx, false_),
+        }
+
+        match self.stop {
+            Some(stop) => {
+                dst_slice_ptr.gep(ctx, |f| f.stop_defined).store(ctx, true_);
+                dst_slice_ptr.gep(ctx, |f| f.stop).store(ctx, stop);
+            }
+            None => dst_slice_ptr.gep(ctx, |f| f.stop_defined).store(ctx, false_),
+        }
+
+        match self.step {
+            Some(step) => {
+                dst_slice_ptr.gep(ctx, |f| f.step_defined).store(ctx, true_);
+                dst_slice_ptr.gep(ctx, |f| f.step).store(ctx, step);
+            }
+            None => dst_slice_ptr.gep(ctx, |f| f.step_defined).store(ctx, false_),
+        }
+    }
+}

nac3core/src/codegen/irrt/numpy/subscript.rs

@@ -0,0 +1,181 @@
+use crate::codegen::{
+    irrt::{
+        error_context::{check_error_context, prepare_error_context, ErrorContext},
+        util::{get_sized_dependent_function_name, FunctionBuilder},
+    },
+    model::*,
+    CodeGenContext, CodeGenerator,
+};
+
+use super::{
+    ndarray::NpArray,
+    slice::{RustUserSlice, SliceIndex, SliceIndexModel, UserSlice},
+};
+
+#[derive(Debug, Clone, Copy)]
+pub struct NDSubscriptFields {
+    pub type_: Field<ByteModel>, // Defined to be uint8_t in IRRT
+    pub data: Field<PointerModel<ByteModel>>,
+}
+
+#[derive(Debug, Clone, Copy, Default)]
+pub struct NDSubscript;
+
+impl<'ctx> IsStruct<'ctx> for NDSubscript {
+    type Fields = NDSubscriptFields;
+
+    fn struct_name(&self) -> &'static str {
+        "NDSubscript"
+    }
+
+    fn build_fields(&self, builder: &mut FieldBuilder<'ctx>) -> Self::Fields {
+        Self::Fields { type_: builder.add_field_auto("type"), data: builder.add_field_auto("data") }
+    }
+}
+
+// An enum variant to store the content
+// and type of an NDSubscript in high level.
+#[derive(Debug, Clone)]
+pub enum RustNDSubscript<'ctx> {
+    Index(SliceIndex<'ctx>),
+    Slice(RustUserSlice<'ctx>),
+}
+
+impl<'ctx> RustNDSubscript<'ctx> {
+    fn irrt_subscript_id(&self) -> u64 {
+        // Defined in IRRT
+        match self {
+            RustNDSubscript::Index(_) => 0,
+            RustNDSubscript::Slice(_) => 1,
+        }
+    }
+
+    fn write_to_ndsubscript(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        dst_ndsubscript_ptr: Pointer<'ctx, StructModel<NDSubscript>>,
+    ) {
+        let byte_model = ByteModel::default();
+        let slice_index_model = SliceIndexModel::default();
+        let user_slice_model = StructModel(UserSlice);
+
+        // Set `dst_ndsubscript_ptr->type`
+        dst_ndsubscript_ptr
+            .gep(ctx, |f| f.type_)
+            .store(ctx, byte_model.constant(ctx.ctx, self.irrt_subscript_id()));
+
+        // Set `dst_ndsubscript_ptr->data`
+        let data = match self {
+            RustNDSubscript::Index(in_index) => {
+                let index_ptr = slice_index_model.alloca(ctx, "index");
+                index_ptr.store(ctx, *in_index);
+                index_ptr.cast_to(ctx, FixedIntModel(Byte), "")
+            }
+            RustNDSubscript::Slice(in_rust_slice) => {
+                let user_slice_ptr = user_slice_model.alloca(ctx, "user_slice");
+                in_rust_slice.write_to_user_slice(ctx, user_slice_ptr);
+                user_slice_ptr.cast_to(ctx, FixedIntModel(Byte), "")
+            }
+        };
+        dst_ndsubscript_ptr.gep(ctx, |f| f.data).store(ctx, data);
+    }
+
+    // Allocate an array of subscripts onto the stack and return its stack pointer
+    pub fn alloca_subscripts(
+        ctx: &CodeGenContext<'ctx, '_>,
+        subscripts: &[RustNDSubscript<'ctx>],
+    ) -> ArraySlice<'ctx, StructModel<NDSubscript>> {
+        let index_model = Int32Model::default();
+
+        let ndsubscript_model = StructModel(NDSubscript);
+        let ndsubscript_array = ndsubscript_model.array_alloca(
+            ctx,
+            index_model.constant(ctx.ctx, subscripts.len() as u64).to_int(),
+            "ndsubscripts",
+        );
+
+        for (i, rust_ndsubscript) in subscripts.iter().enumerate() {
+            let ndsubscript_ptr = ndsubscript_array.ix_unchecked(
+                ctx,
+                index_model.constant(ctx.ctx, i as u64).to_int(),
+                "",
+            );
+            rust_ndsubscript.write_to_ndsubscript(ctx, ndsubscript_ptr);
+        }
+
+        ndsubscript_array
+    }
+
+    #[must_use]
+    pub fn deduce_ndims_after_slicing(slices: &[RustNDSubscript], original_ndims: i32) -> i32 {
+        let mut final_ndims: i32 = original_ndims;
+        for slice in slices {
+            match slice {
+                RustNDSubscript::Index(_) => {
+                    // Index subscripts demotes the rank by 1
+                    final_ndims -= 1;
+                }
+                RustNDSubscript::Slice(_) => {
+                    // Nothing
+                }
+            }
+        }
+        final_ndims
+    }
+}
+
+pub fn call_nac3_ndarray_subscript_deduce_ndims_after_slicing<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    sizet: IntModel<'ctx>,
+    ndims: Int<'ctx>,
+    num_ndsubscripts: Int<'ctx>,
+    ndsubscripts: Pointer<'ctx, StructModel<NDSubscript>>,
+) -> Int<'ctx> {
+    let result = sizet.alloca(ctx, "result");
+
+    let errctx_ptr = prepare_error_context(ctx);
+    FunctionBuilder::begin(
+        ctx,
+        &get_sized_dependent_function_name(
+            sizet,
+            "__nac3_ndarray_subscript_deduce_ndims_after_slicing",
+        ),
+    )
+    .arg("errctx", PointerModel(StructModel(ErrorContext)), errctx_ptr)
+    .arg("result", PointerModel(sizet), result)
+    .arg("ndims", sizet, ndims)
+    .arg("num_ndsubscripts", sizet, num_ndsubscripts)
+    .arg("ndsubscripts", PointerModel(StructModel(NDSubscript)), ndsubscripts)
+    .returning_void();
+    check_error_context(generator, ctx, errctx_ptr);
+
+    result.load(ctx, "final_ndims")
+}
+
+pub fn call_nac3_ndarray_subscript<'ctx, G: CodeGenerator + ?Sized>(
+    generator: &mut G,
+    ctx: &mut CodeGenContext<'ctx, '_>,
+    num_subscripts: SliceIndex<'ctx>,
+    subscripts: Pointer<'ctx, StructModel<NDSubscript>>,
+    src_ndarray: Pointer<'ctx, StructModel<NpArray<'ctx>>>,
+    dst_ndarray: Pointer<'ctx, StructModel<NpArray<'ctx>>>,
+) {
+    let sizet = src_ndarray.element.0.sizet;
+    assert!(sizet.same_as(dst_ndarray.element.0.sizet)); // SizeT of src_ndarray and dst_ndarray must match
+
+    let errctx_ptr = prepare_error_context(ctx);
+
+    FunctionBuilder::begin(
+        ctx,
+        &get_sized_dependent_function_name(sizet, "__nac3_ndarray_subscript"),
+    )
+    .arg("errctx", PointerModel(StructModel(ErrorContext)), errctx_ptr)
+    .arg("num_subscripts", SliceIndexModel::default(), num_subscripts)
+    .arg("subscripts", PointerModel(StructModel(NDSubscript)), subscripts)
+    .arg("src_ndarray", PointerModel(StructModel(NpArray { sizet })), src_ndarray)
+    .arg("dst_ndarray", PointerModel(StructModel(NpArray { sizet })), dst_ndarray)
+    .returning_void();
+
+    check_error_context(generator, ctx, errctx_ptr);
+}

nac3core/src/codegen/irrt/range.rs

@@ -1,56 +0,0 @@
-use inkwell::{
-    values::{BasicValueEnum, CallSiteValue, IntValue},
-    IntPredicate,
-};
-use itertools::Either;
-
-use crate::codegen::{CodeGenContext, CodeGenerator};
-
-/// Invokes the `__nac3_range_slice_len` in IRRT.
-///
-/// - `start`: The `i32` start value for the slice.
-/// - `end`: The `i32` end value for the slice.
-/// - `step`: The `i32` step value for the slice.
-///
-/// Returns an `i32` value of the length of the slice.
-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 llvm_i32 = ctx.ctx.i32_type();
-
-    assert_eq!(start.get_type(), llvm_i32);
-    assert_eq!(end.get_type(), llvm_i32);
-    assert_eq!(step.get_type(), llvm_i32);
-
-    let len_func = ctx.module.get_function(SYMBOL).unwrap_or_else(|| {
-        let fn_t = llvm_i32.fn_type(&[llvm_i32.into(), llvm_i32.into(), llvm_i32.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/slice.rs

@@ -1,39 +0,0 @@
-use inkwell::values::{BasicValueEnum, CallSiteValue, IntValue};
-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(),
-    ))
-}

nac3core/src/codegen/irrt/string.rs

@@ -1,45 +0,0 @@
-use inkwell::values::{BasicValueEnum, CallSiteValue, IntValue, PointerValue};
-use itertools::Either;
-
-use super::get_usize_dependent_function_name;
-use crate::codegen::CodeGenContext;
-
-/// Generates a call to string equality comparison. Returns an `i1` representing whether the strings are equal.
-pub fn call_string_eq<'ctx>(
-    ctx: &CodeGenContext<'ctx, '_>,
-    str1_ptr: PointerValue<'ctx>,
-    str1_len: IntValue<'ctx>,
-    str2_ptr: PointerValue<'ctx>,
-    str2_len: IntValue<'ctx>,
-) -> IntValue<'ctx> {
-    let llvm_i1 = ctx.ctx.bool_type();
-
-    let func_name = get_usize_dependent_function_name(ctx, "nac3_str_eq");
-
-    let func = ctx.module.get_function(&func_name).unwrap_or_else(|| {
-        ctx.module.add_function(
-            &func_name,
-            llvm_i1.fn_type(
-                &[
-                    str1_ptr.get_type().into(),
-                    str1_len.get_type().into(),
-                    str2_ptr.get_type().into(),
-                    str2_len.get_type().into(),
-                ],
-                false,
-            ),
-            None,
-        )
-    });
-
-    ctx.builder
-        .build_call(
-            func,
-            &[str1_ptr.into(), str1_len.into(), str2_ptr.into(), str2_len.into()],
-            "str_eq_call",
-        )
-        .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

@@ -0,0 +1,26 @@
+#[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");
+        }
+    }
+}