WIP

core/irrt: add c-string utils
core: refactor to lift `struct SizeVariant` to /util.rs
2024-07-18 16:07:47 +08:00 · 2024-07-17 17:50:55 +08:00 · 2024-07-17 17:50:49 +08:00 · 2024-07-17 17:50:39 +08:00 · 2024-07-17 17:50:31 +08:00 · 2024-07-17 17:50:23 +08:00
34 changed files with 1802 additions and 185 deletions
--- a/flake.nix
+++ b/flake.nix
@ -13,6 +13,7 @@
          ''
          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
          '';
        nac3artiq = pkgs.python3Packages.toPythonModule (
@ -23,6 +24,7 @@
            cargoLock = {
              lockFile = ./Cargo.lock;
            };
+            cargoTestFlags = [ "--features" "test" ];
            passthru.cargoLock = cargoLock;
            nativeBuildInputs = [ pkgs.python3 pkgs.llvmPackages_14.clang llvm-tools-irrt pkgs.llvmPackages_14.llvm.out llvm-nac3 ];
            buildInputs = [ pkgs.python3 llvm-nac3 ];
--- a/nac3core/Cargo.toml
+++ b/nac3core/Cargo.toml
@ -1,3 +1,6 @@
+[features]
+test = []
+
 [package]
 name = "nac3core"
 version = "0.1.0"
--- a/nac3core/build.rs
+++ b/nac3core/build.rs
@ -3,20 +3,34 @@ use std::{
    env,
    fs::File,
    io::Write,
-    path::Path,
+    path::{Path, PathBuf},
    process::{Command, Stdio},
 };

-fn main() {
-    const FILE: &str = "src/codegen/irrt/irrt.cpp";
+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 get_out_dir() -> PathBuf {
+    PathBuf::from(env::var("OUT_DIR").unwrap())
+}
+
+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 irrt_cpp_path = irrt_dir.join("irrt.cpp");
    let flags: &[&str] = &[
        "--target=wasm32",
-        FILE,
        "-x",
        "c++",
        "-fno-discard-value-names",
@ -31,15 +45,19 @@ fn main() {
        "-S",
        "-Wall",
        "-Wextra",
+        "-Werror=return-type",
        "-o",
        "-",
+        "-I",
+        irrt_dir.to_str().unwrap(),
+        irrt_cpp_path.to_str().unwrap(),
    ];

-    println!("cargo:rerun-if-changed={FILE}");
-    let out_dir = env::var("OUT_DIR").unwrap();
-    let out_path = Path::new(&out_dir);
+    // Tell Cargo to rerun if any file under `irrt_dir` (recursive) changes
+    println!("cargo:rerun-if-changed={}", irrt_dir.to_str().unwrap());

-    let output = Command::new("clang-irrt")
+    // Compile IRRT and capture the LLVM IR output
+    let output = Command::new(CMD_IRRT_CLANG)
        .args(flags)
        .output()
        .map(|o| {
@ -52,7 +70,17 @@ fn main() {
    let output = std::str::from_utf8(&output.stdout).unwrap().replace("\r\n", "\n");
    let mut filtered_output = String::with_capacity(output.len());

-    let regex_filter = Regex::new(r"(?ms:^define.*?\}$)|(?m:^declare.*?$)").unwrap();
+    // Filter out irrelevant IR
+    //
+    // Regex:
+    // - `(?ms:^define.*?\}$)` captures LLVM `define` blocks
+    // - `(?m:^declare.*?$)` captures LLVM `declare` lines
+    // - `(?m:^%.+?=\s*type\s*\{.+?\}$)` captures LLVM `type` declarations
+    // - `(?m:^@.+?=.+$)` captures global constants
+    let regex_filter = Regex::new(
+        r"(?ms:^define.*?\}$)|(?m:^declare.*?$)|(?m:^%.+?=\s*type\s*\{.+?\}$)|(?m:^@.+?=.+$)",
+    )
+    .unwrap();
    for f in regex_filter.captures_iter(&output) {
        assert_eq!(f.len(), 1);
        filtered_output.push_str(&f[0]);
@ -63,20 +91,71 @@ fn main() {
        .unwrap()
        .replace_all(&filtered_output, "");

-    println!("cargo:rerun-if-env-changed=DEBUG_DUMP_IRRT");
-    if env::var("DEBUG_DUMP_IRRT").is_ok() {
-        let mut file = File::create(out_path.join("irrt.ll")).unwrap();
+    // For debugging
+    // Doing `DEBUG_DUMP_IRRT=1 cargo build -p nac3core` dumps the LLVM IR generated
+    const DEBUG_DUMP_IRRT: &str = "DEBUG_DUMP_IRRT";
+    println!("cargo:rerun-if-env-changed={DEBUG_DUMP_IRRT}");
+    if env::var(DEBUG_DUMP_IRRT).is_ok() {
+        let mut file = File::create(out_dir.join("irrt.ll")).unwrap();
        file.write_all(output.as_bytes()).unwrap();
-        let mut file = File::create(out_path.join("irrt-filtered.ll")).unwrap();
+
+        let mut file = File::create(out_dir.join("irrt-filtered.ll")).unwrap();
        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_path.join("irrt.bc"))
+        .arg(out_dir.join("irrt.bc"))
        .spawn()
        .unwrap();
    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()`
+        "-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();
+    }
+}
--- a/nac3core/irrt/irrt.cpp
+++ b/nac3core/irrt/irrt.cpp
@ -0,0 +1,9 @@
+#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.
+*/
--- a/nac3core/src/codegen/irrt/irrt.cpp
+++ b/nac3core/src/codegen/irrt/irrt.cpp
@ -1,27 +1,17 @@
-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);
+#pragma once
+
+#include <irrt/int_defs.hpp>
+#include <irrt/utils.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`.
+// The type of an index or a value describing the length of a
+// range/slice is always `int32_t`.
 using SliceIndex = int32_t;

 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;
-}
-
-// adapted from GNU Scientific Library: https://git.savannah.gnu.org/cgit/gsl.git/tree/sys/pow_int.c
+// 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) {
@ -38,12 +28,7 @@ T __nac3_int_exp_impl(T base, T exp) {
 }

 template <typename SizeT>
-SizeT __nac3_ndarray_calc_size_impl(
-    const SizeT* list_data,
-    SizeT list_len,
-    SizeT begin_idx,
-    SizeT end_idx
-) {
+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;
@ -56,12 +41,7 @@ SizeT __nac3_ndarray_calc_size_impl(
 }

 template <typename SizeT>
-void __nac3_ndarray_calc_nd_indices_impl(
-    SizeT index,
-    const SizeT* dims,
-    SizeT num_dims,
-    NDIndex* idxs
-) {
+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;
@ -72,12 +52,7 @@ void __nac3_ndarray_calc_nd_indices_impl(
 }

 template <typename SizeT>
-SizeT __nac3_ndarray_flatten_index_impl(
-    const SizeT* dims,
-    SizeT num_dims,
-    const NDIndex* indices,
-    SizeT num_indices
-) {
+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) {
@ -94,17 +69,14 @@ SizeT __nac3_ndarray_flatten_index_impl(

 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
+    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) {
@ -125,10 +97,7 @@ void __nac3_ndarray_calc_broadcast_impl(

 template <typename SizeT>
 void __nac3_ndarray_calc_broadcast_idx_impl(
-    const SizeT* src_dims,
-    SizeT src_ndims,
-    const NDIndex* in_idx,
-    NDIndex* out_idx
+    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;
@ -139,14 +108,14 @@ void __nac3_ndarray_calc_broadcast_idx_impl(

 extern "C" {
 #define DEF_nac3_int_exp_(T)                   \
-    T __nac3_int_exp_##T(T base, T exp) {\
-        return __nac3_int_exp_impl(base, exp);\
+    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)
+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) {
@ -160,11 +129,7 @@ SliceIndex __nac3_slice_index_bound(SliceIndex i, const SliceIndex len) {
    return i;
 }

-SliceIndex __nac3_range_slice_len(
-    const SliceIndex start,
-    const SliceIndex end,
-    const SliceIndex step
-) {
+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;
@ -180,7 +145,8 @@ SliceIndex __nac3_range_slice_len(
 // - 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)
+// 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,
@ -194,18 +160,18 @@ SliceIndex __nac3_list_slice_assign_var_size(
    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 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
-            );
+            __builtin_memmove(dest_arr + dest_start * size, src_arr + src_start * size, src_len * size);
        }
        if (dest_len > 0) {
            /* dropping */
@ -219,23 +185,17 @@ SliceIndex __nac3_list_slice_assign_var_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)
+    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));
+        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 (; (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);
@ -244,7 +204,8 @@ SliceIndex __nac3_list_slice_assign_var_size(
        } 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 */
+            /* memcpy for var size, cannot overlap after previous
+             * alloca */
            __builtin_memcpy(dest_arr + dest_ind * size, src_arr + src_ind * size, size);
        }
    }
@ -254,7 +215,7 @@ SliceIndex __nac3_list_slice_assign_var_size(
        __builtin_memmove(
            dest_arr + dest_ind * size,
            dest_arr + (dest_end + 1) * size,
-            (dest_arr_len - dest_end - 1) * size
+            (dest_arr_len - dest_end - 1) * size + size + size + size
        );
        return dest_arr_len - (dest_end - dest_ind) - 1;
    }
@ -320,94 +281,53 @@ double __nac3_j0(double x) {
    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
-) {
+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
-) {
+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
-) {
+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
-) {
+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
-) {
+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
-) {
+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
+    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
+    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
+    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
+    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);
 }
--- a/nac3core/irrt/irrt/error_context.hpp
+++ b/nac3core/irrt/irrt/error_context.hpp
@ -0,0 +1,87 @@
+#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
+    ErrorIds* error_ids;
+
+    // Error thrown by IRRT
+    ErrorId error_id;
+    const char* message_template;  // MUST BE `&'static`
+    uint64_t param1;
+    uint64_t param2;
+    uint64_t param3;
+
+    void initialize(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, uint64_t param1 = 0, uint64_t param2 = 0, uint64_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); }
+
+    /// Get a nac3core-understanding `Str<SizeT>` that containing
+    /// the error 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);
+    }
+};
+}  // namespace
+
+extern "C" {
+void __nac3_error_context_initialize(ErrorContext* errctx, 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);
+}
+
+// Used for testing
+void __nac3_error_dummy_raise(ErrorContext* errctx) {
+    errctx->set_error(errctx->error_ids->runtime_error, "Error thrown from __nac3_error_dummy_raise");
+}
+}
--- a/nac3core/irrt/irrt/int_defs.hpp
+++ b/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
+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);
+#endif
--- a/nac3core/irrt/irrt/utils.hpp
+++ b/nac3core/irrt/irrt/utils.hpp
@ -0,0 +1,77 @@
+#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;
+}
+
+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();
+}
+}  // namespace cstr_utils
+}  // namespace
--- a/nac3core/irrt/irrt_everything.hpp
+++ b/nac3core/irrt/irrt_everything.hpp
@ -0,0 +1,6 @@
+#pragma once
+
+#include <irrt/core.hpp>
+#include <irrt/int_defs.hpp>
+#include <irrt/utils.hpp>
+#include <irrt/error_context.hpp>
--- a/nac3core/irrt/irrt_test.cpp
+++ b/nac3core/irrt/irrt_test.cpp
@ -0,0 +1,13 @@
+// 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 <test/test_core.hpp>
+
+int main() {
+    test::core::run();
+    return 0;
+}
--- a/nac3core/irrt/test/includes.hpp
+++ b/nac3core/irrt/test/includes.hpp
@ -0,0 +1,12 @@
+#pragma once
+
+#include <cstdint>
+#include <cstdio>
+#include <cstdlib>
+
+#include <test/util.hpp>
+#include <irrt_everything.hpp>
+
+/*
+    Include this header for every test_*.cpp
+*/
--- a/nac3core/irrt/test/test_core.hpp
+++ b/nac3core/irrt/test/test_core.hpp
@ -0,0 +1,16 @@
+#pragma once
+
+#include <test/includes.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();
+}
+}}  // namespace test::core
--- a/nac3core/irrt/test/util.hpp
+++ b/nac3core/irrt/test/util.hpp
@ -0,0 +1,114 @@
+#pragma once
+
+#include <cstdlib>
+#include <cstdio>
+
+template <class T>
+void print_value(const T& value) {}
+
+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 __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);
+}
+
+template <typename T>
+void debug_print_array(int len, const T* as) {
+    printf("[");
+    for (int i = 0; i < len; i++) {
+        if (i != 0)
+            printf(", ");
+        print_value(as[i]);
+    }
+    printf("]");
+}
+
+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 = ");
+        debug_print_array(len, expected);
+        printf("\n");
+        printf("   Got = ");
+        debug_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/nac3core/src/codegen/expr.rs
+++ b/nac3core/src/codegen/expr.rs
@ -576,6 +576,21 @@ impl<'ctx, 'a> CodeGenContext<'ctx, 'a> {
        params: [Option<IntValue<'ctx>>; 3],
        loc: Location,
    ) {
+        let error_id = self.resolver.get_string_id(name);
+        let error_id = self.ctx.i32_type().const_int(error_id as u64, false);
+        self.raise_exn_by_id(generator, error_id, msg, params, loc);
+    }
+
+    pub fn raise_exn_by_id<G: CodeGenerator + ?Sized>(
+        &mut self,
+        generator: &mut G,
+        error_id: IntValue<'ctx>,
+        msg: BasicValueEnum<'ctx>,
+        params: [Option<IntValue<'ctx>>; 3],
+        loc: Location,
+    ) {
+        assert_eq!(error_id.get_type().get_bit_width(), 32);
+
        let zelf = if let Some(exception_val) = self.exception_val {
            exception_val
        } else {
@ -587,9 +602,6 @@ impl<'ctx, 'a> CodeGenContext<'ctx, 'a> {
        let int32 = self.ctx.i32_type();
        let zero = int32.const_zero();
        unsafe {
-            let id_ptr = self.builder.build_in_bounds_gep(zelf, &[zero, zero], "exn.id").unwrap();
-            let id = self.resolver.get_string_id(name);
-            self.builder.build_store(id_ptr, int32.const_int(id as u64, false)).unwrap();
            let ptr = self
                .builder
                .build_in_bounds_gep(zelf, &[zero, int32.const_int(5, false)], "exn.msg")
@ -652,6 +664,32 @@ impl<'ctx, 'a> CodeGenContext<'ctx, 'a> {
        self.raise_exn(generator, err_name, err_msg, params, loc);
        self.builder.position_at_end(then_block);
    }
+
+    pub fn make_assert_impl_by_id<G: CodeGenerator + ?Sized>(
+        &mut self,
+        generator: &mut G,
+        cond: IntValue<'ctx>,
+        err_id: IntValue<'ctx>,
+        err_msg: BasicValueEnum<'ctx>,
+        params: [Option<IntValue<'ctx>>; 3],
+        loc: Location,
+    ) {
+        let i1 = self.ctx.bool_type();
+        let i1_true = i1.const_all_ones();
+        // we assume that the condition is most probably true, so the normal path is the most
+        // probable path
+        // even if this assumption is violated, it does not matter as exception unwinding is
+        // slow anyway...
+        let cond = call_expect(self, cond, i1_true, Some("expect"));
+        let current_bb = self.builder.get_insert_block().unwrap();
+        let current_fun = current_bb.get_parent().unwrap();
+        let then_block = self.ctx.insert_basic_block_after(current_bb, "succ");
+        let exn_block = self.ctx.append_basic_block(current_fun, "fail");
+        self.builder.build_conditional_branch(cond, then_block, exn_block).unwrap();
+        self.builder.position_at_end(exn_block);
+        self.raise_exn_by_id(generator, err_id, err_msg, params, loc);
+        self.builder.position_at_end(then_block);
+    }
 }

 /// See [`CodeGenerator::gen_constructor`].
--- a/nac3core/src/codegen/irrt/error_context.rs
+++ b/nac3core/src/codegen/irrt/error_context.rs
@ -0,0 +1,185 @@
+use crate::codegen::{model::*, CodeGenContext, CodeGenerator};
+
+use super::{string::Str, util::get_sized_dependent_function_name};
+
+/// The [`IntModel`] of nac3core's error ID.
+///
+/// It is always [`Int32`].
+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>>,
+}
+
+/// Corresponds to IRRT's `struct ErrorIds`
+#[derive(Debug, Clone, Copy, Default)]
+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_ids: Field<PointerModel<StructModel<ErrorIds>>>,
+    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>>,
+}
+
+/// Corresponds to IRRT's `struct ErrorContext`
+#[derive(Debug, Clone, Copy, Default)]
+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_ids: builder.add_field_auto("error_ids"),
+            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();
+}
+
+/// Setup a [`ErrorContext`] that could
+/// be passed to IRRT functions taking in a `ErrorContext* errctx`
+/// for error reporting purposes.
+///
+/// Also see: [`check_error_context`]
+pub fn setup_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
+}
+
+/// Check a [`ErrorContext`] to see
+/// if it contains error.
+///
+/// If there is an error, an LLVM exception will be raised at runtime.
+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));
+
+    // Does ErrorContext contain an error?
+    let has_error = call_nac3_error_context_has_no_error(ctx, errctx_ptr);
+
+    // Get the error message (doesn't matter even if there's actually no error)
+    let pstr = StructModel(Str { sizet }).alloca(ctx, "error_str");
+    call_nac3_error_context_get_error_str(sizet, ctx, errctx_ptr, pstr);
+
+    // Load all the values for `ctx.make_assert_impl_by_id`
+    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");
+
+    // Make assert
+    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 = setup_error_context(ctx);
+    FunctionBuilder::begin(ctx, "__nac3_error_dummy_raise")
+        .arg("errctx", PointerModel(StructModel(ErrorContext)), errctx)
+        .returning_void();
+    check_error_context(generator, ctx, errctx);
+}
--- a/nac3core/src/codegen/irrt/mod.rs
+++ b/nac3core/src/codegen/irrt/mod.rs
@ -1,5 +1,10 @@
 use crate::typecheck::typedef::Type;

+pub mod error_context;
+pub mod string;
+mod test;
+pub mod util;
+
 use super::{
    classes::{
        ArrayLikeIndexer, ArrayLikeValue, ArraySliceValue, ListValue, NDArrayValue,
--- a/nac3core/src/codegen/irrt/string.rs
+++ b/nac3core/src/codegen/irrt/string.rs
@ -0,0 +1,34 @@
+use crate::codegen::model::*;
+
+pub struct StrFields<'ctx> {
+    /// Pointer to the string. Does not have to be null-terminated.
+    pub content: Field<PointerModel<FixedIntModel<Byte>>>,
+    /// Number of bytes this string occupies in space.
+    ///
+    /// The [`IntModel`] matches [`Str::sizet`].
+    pub length: Field<IntModel<'ctx>>,
+}
+
+/// Corresponds to IRRT's `struct Str`
+///
+/// nac3core's LLVM representation of a string in memory
+#[derive(Debug, Clone, Copy)]
+pub struct Str<'ctx> {
+    /// The `SizeT` type of this string.
+    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),
+        }
+    }
+}
--- a/nac3core/src/codegen/irrt/test.rs
+++ b/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");
+        }
+    }
+}
--- a/nac3core/src/codegen/irrt/util.rs
+++ b/nac3core/src/codegen/irrt/util.rs
@ -0,0 +1,79 @@
+use inkwell::{
+    types::{BasicMetadataTypeEnum, BasicType, IntType},
+    values::{AnyValue, BasicMetadataValueEnum},
+};
+
+use crate::{
+    codegen::{model::*, CodeGenContext},
+    util::SizeVariant,
+};
+
+fn get_size_variant(ty: IntType) -> SizeVariant {
+    match ty.get_bit_width() {
+        32 => SizeVariant::Bits32,
+        64 => SizeVariant::Bits64,
+        _ => unreachable!("Unsupported int type bit width {}", ty.get_bit_width()),
+    }
+}
+
+#[must_use]
+pub fn get_sized_dependent_function_name(ty: IntModel, fn_name: &str) -> String {
+    let mut fn_name = fn_name.to_owned();
+    match get_size_variant(ty.0) {
+        SizeVariant::Bits32 => {
+            // Do nothing, `fn_name` already has the correct name
+        }
+        SizeVariant::Bits64 => {
+            // Append "64", this is the naming convention
+            fn_name.push_str("64");
+        }
+    }
+    fn_name
+}
+
+// TODO: Variadic argument?
+pub struct FunctionBuilder<'ctx, 'a> {
+    ctx: &'a CodeGenContext<'ctx, 'a>,
+    fn_name: &'a str,
+    arguments: Vec<(BasicMetadataTypeEnum<'ctx>, BasicMetadataValueEnum<'ctx>)>,
+}
+
+impl<'ctx, 'a> FunctionBuilder<'ctx, 'a> {
+    pub fn begin(ctx: &'a CodeGenContext<'ctx, 'a>, fn_name: &'a str) -> Self {
+        FunctionBuilder { ctx, fn_name, arguments: Vec::new() }
+    }
+
+    // The name is for self-documentation
+    #[must_use]
+    pub fn arg<M: Model<'ctx>>(mut self, _name: &'static str, model: M, value: M::Value) -> Self {
+        self.arguments
+            .push((model.get_llvm_type(self.ctx.ctx).into(), value.get_llvm_value().into()));
+        self
+    }
+
+    pub fn returning<M: Model<'ctx>>(self, name: &'static str, return_model: M) -> M::Value {
+        let (param_tys, param_vals): (Vec<_>, Vec<_>) = self.arguments.into_iter().unzip();
+
+        let function = self.ctx.module.get_function(self.fn_name).unwrap_or_else(|| {
+            let return_type = return_model.get_llvm_type(self.ctx.ctx);
+            let fn_type = return_type.fn_type(&param_tys, false);
+            self.ctx.module.add_function(self.fn_name, fn_type, None)
+        });
+
+        let ret = self.ctx.builder.build_call(function, &param_vals, name).unwrap();
+        return_model.review(self.ctx.ctx, ret.as_any_value_enum())
+    }
+
+    // TODO: Code duplication, but otherwise returning<S: Optic<'ctx>> cannot resolve S if return_optic = None
+    pub fn returning_void(self) {
+        let (param_tys, param_vals): (Vec<_>, Vec<_>) = self.arguments.into_iter().unzip();
+
+        let function = self.ctx.module.get_function(self.fn_name).unwrap_or_else(|| {
+            let return_type = self.ctx.ctx.void_type();
+            let fn_type = return_type.fn_type(&param_tys, false);
+            self.ctx.module.add_function(self.fn_name, fn_type, None)
+        });
+
+        self.ctx.builder.build_call(function, &param_vals, "").unwrap();
+    }
+}
--- a/nac3core/src/codegen/mod.rs
+++ b/nac3core/src/codegen/mod.rs
@ -23,7 +23,9 @@ use inkwell::{
    values::{BasicValueEnum, FunctionValue, IntValue, PhiValue, PointerValue},
    AddressSpace, IntPredicate, OptimizationLevel,
 };
+use irrt::string::Str;
 use itertools::Itertools;
+use model::*;
 use nac3parser::ast::{Location, Stmt, StrRef};
 use parking_lot::{Condvar, Mutex};
 use std::collections::{HashMap, HashSet};
@ -41,6 +43,7 @@ pub mod extern_fns;
 mod generator;
 pub mod irrt;
 pub mod llvm_intrinsics;
+pub mod model;
 pub mod numpy;
 pub mod stmt;

@ -654,19 +657,8 @@ pub fn gen_func_impl<
        (primitives.float, context.f64_type().into()),
        (primitives.bool, context.i8_type().into()),
        (primitives.str, {
-            let name = "str";
-            match module.get_struct_type(name) {
-                None => {
-                    let str_type = context.opaque_struct_type("str");
-                    let fields = [
-                        context.i8_type().ptr_type(AddressSpace::default()).into(),
-                        generator.get_size_type(context).into(),
-                    ];
-                    str_type.set_body(&fields, false);
-                    str_type.into()
-                }
-                Some(t) => t.as_basic_type_enum(),
-            }
+            let sizet = IntModel(generator.get_size_type(context));
+            StructModel(Str { sizet }).get_llvm_type(context)
        }),
        (primitives.range, RangeType::new(context).as_base_type().into()),
        (primitives.exception, {
--- a/nac3core/src/codegen/model/core.rs
+++ b/nac3core/src/codegen/model/core.rs
@ -0,0 +1,118 @@
+use inkwell::{
+    context::Context,
+    types::{AnyTypeEnum, BasicTypeEnum},
+    values::{AnyValue, AnyValueEnum, BasicValueEnum, PointerValue},
+};
+
+use crate::codegen::{CodeGenContext, CodeGenerator};
+
+use super::{slice::ArraySlice, Int, Pointer};
+
+/// A value that belongs to/produced by a [`Model<'ctx>`]
+pub trait ModelValue<'ctx>: Clone + Copy {
+    fn get_llvm_value(&self) -> BasicValueEnum<'ctx>;
+}
+
+// Should have been within [`Model<'ctx>`],
+// but rust object safety requirements made it necessary to
+// split this interface out
+pub trait CanCheckLLVMType<'ctx> {
+    /// Check if `scrutinee` matches the same LLVM type of this [`Model<'ctx>`].
+    ///
+    /// If they don't not match, a human-readable error message is returned.
+    fn check_llvm_type(
+        &self,
+        ctx: &'ctx Context,
+        scrutinee: AnyTypeEnum<'ctx>,
+    ) -> Result<(), String>;
+}
+
+/// A [`Model`] is a type-safe concrete representation of a complex LLVM type.
+pub trait Model<'ctx>: Clone + Copy + CanCheckLLVMType<'ctx> + Sized {
+    /// The values that inhabit this [`Model<'ctx>`].
+    ///
+    /// ...that is the type of wrapper that wraps the LLVM values that inhabit [`Model<'ctx>::get_llvm_type()`].
+    type Value: ModelValue<'ctx>;
+
+    /// Get the [`BasicTypeEnum<'ctx>`] this [`Model<'ctx>`] is representing.
+    fn get_llvm_type(&self, ctx: &'ctx Context) -> BasicTypeEnum<'ctx>;
+
+    /// Cast an [`AnyValueEnum<'ctx>`] into [`Self::Value`].
+    ///
+    /// Panics if `value` cannot pass [`CanCheckLLVMType::check_llvm_type()`].
+    fn review(&self, ctx: &'ctx Context, value: AnyValueEnum<'ctx>) -> Self::Value;
+
+    /// Check if [`Self::Value`] has the correct type described by this [`Model<'ctx>`]
+    ///
+    /// For example:
+    /// ```ignore
+    /// let ctx: &CodeGenContext<'ctx, '_>;
+    /// let my_i32 = IntModel(ctx.ctx.i32_type());
+    /// let my_i64 = IntModel(ctx.ctx.i64_type());
+    /// let value1 = my_i32.constant(3);
+    /// let value2 = my_i64.constant(3);
+    /// // Both value1 and value2 have type `IntModel<'ctx>`!
+    /// // There is no type constraints to tell which value has what int type.
+    /// my_i32.check(value1); // ok
+    /// my_i64.check(value2); // ok
+    ///
+    /// my_i32.check(value2); // PANIC
+    /// my_i64.check(value1); // PANIC
+    /// ```
+    fn check(&self, ctx: &'ctx Context, value: Self::Value) {
+        self.review(ctx, value.get_llvm_value().as_any_value_enum());
+    }
+
+    /// Build an instruction to allocate a value of [`Model::get_llvm_type`].
+    fn alloca(&self, ctx: &CodeGenContext<'ctx, '_>, name: &str) -> Pointer<'ctx, Self> {
+        Pointer {
+            element: *self,
+            value: ctx.builder.build_alloca(self.get_llvm_type(ctx.ctx), name).unwrap(),
+        }
+    }
+
+    /// Build an instruction to allocate an array of [`Model::get_llvm_type`].
+    fn array_alloca(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        count: Int<'ctx>,
+        name: &str,
+    ) -> ArraySlice<'ctx, Self> {
+        ArraySlice {
+            num_elements: count,
+            pointer: Pointer {
+                element: *self,
+                value: ctx
+                    .builder
+                    .build_array_alloca(self.get_llvm_type(ctx.ctx), count.0, name)
+                    .unwrap(),
+            },
+        }
+    }
+
+    /// Do [`CodeGenerator::gen_var_alloc`] with the LLVM type of this [`Model<'ctx>`].
+    fn var_alloc<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        name: Option<&str>,
+    ) -> Result<Pointer<'ctx, Self>, String> {
+        let value = generator.gen_var_alloc(ctx, self.get_llvm_type(ctx.ctx), name)?;
+        Ok(Pointer { element: *self, value })
+    }
+
+    /// Do [`CodeGenerator::gen_array_var_alloc`] with the LLVM type of this [`Model<'ctx>`].
+    fn array_var_alloc<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        size: Int<'ctx>,
+        name: Option<&'ctx str>,
+    ) -> Result<Pointer<'ctx, Self>, String> {
+        let slice =
+            generator.gen_array_var_alloc(ctx, self.get_llvm_type(ctx.ctx), size.0, name)?;
+        let ptr = PointerValue::from(slice); // TODO: Remove ArraySliceValue
+
+        Ok(Pointer { element: *self, value: ptr })
+    }
+}
--- a/nac3core/src/codegen/model/fixed_int.rs
+++ b/nac3core/src/codegen/model/fixed_int.rs
@ -0,0 +1,159 @@
+use inkwell::{
+    context::Context,
+    types::{AnyTypeEnum, BasicType, BasicTypeEnum, IntType},
+    values::{AnyValueEnum, BasicValue, BasicValueEnum, IntValue},
+};
+
+use crate::codegen::CodeGenContext;
+
+use super::{
+    core::*,
+    int_util::{check_int_llvm_type, review_int_llvm_value},
+    Int, IntModel,
+};
+
+/// A marker trait to mark singleton struct that describes a particular fixed integer type.
+/// See [`Bool`], [`Byte`], [`Int32`], etc.
+///
+/// The [`Default`] trait is to enable auto-derivations for utilities like
+/// [`FieldBuilder::add_field_auto`]
+pub trait IsFixedInt: Clone + Copy + Default {
+    fn get_int_type(ctx: &Context) -> IntType<'_>;
+    fn get_bit_width() -> u32; // This is required, instead of only relying on get_int_type
+
+    fn constant<'ctx>(&self, ctx: &'ctx Context, value: u64) -> FixedInt<'ctx, Self> {
+        FixedInt { int: *self, value: Self::get_int_type(ctx).const_int(value, false) }
+    }
+}
+
+// Some pre-defined fixed integers
+
+#[derive(Debug, Clone, Copy, Default)]
+pub struct Bool;
+pub type BoolModel = FixedIntModel<Bool>;
+
+impl IsFixedInt for Bool {
+    fn get_int_type(ctx: &Context) -> IntType<'_> {
+        ctx.bool_type()
+    }
+
+    fn get_bit_width() -> u32 {
+        1
+    }
+}
+
+#[derive(Debug, Clone, Copy, Default)]
+pub struct Byte;
+pub type ByteModel = FixedIntModel<Byte>;
+
+impl IsFixedInt for Byte {
+    fn get_int_type(ctx: &Context) -> IntType<'_> {
+        ctx.i8_type()
+    }
+
+    fn get_bit_width() -> u32 {
+        8
+    }
+}
+
+#[derive(Debug, Clone, Copy, Default)]
+pub struct Int32;
+pub type Int32Model = FixedIntModel<Int32>;
+
+impl IsFixedInt for Int32 {
+    fn get_int_type(ctx: &Context) -> IntType<'_> {
+        ctx.i32_type()
+    }
+
+    fn get_bit_width() -> u32 {
+        32
+    }
+}
+
+#[derive(Debug, Clone, Copy, Default)]
+pub struct Int64;
+pub type Int64Model = FixedIntModel<Int64>;
+
+impl IsFixedInt for Int64 {
+    fn get_int_type(ctx: &Context) -> IntType<'_> {
+        ctx.i64_type()
+    }
+
+    fn get_bit_width() -> u32 {
+        64
+    }
+}
+
+/// A model representing a compile-time known [`IntType<'ctx>`].
+///
+/// Also see [`IntModel`], which is less constrained than [`FixedIntModel`],
+/// but enables one to handle [`IntType<'ctx>`] that could be dynamic
+#[derive(Debug, Clone, Copy, Default)]
+pub struct FixedIntModel<T>(pub T);
+
+// FixedIntModel's implementation
+
+impl<'ctx, T: IsFixedInt> CanCheckLLVMType<'ctx> for FixedIntModel<T> {
+    fn check_llvm_type(
+        &self,
+        ctx: &'ctx Context,
+        scrutinee: AnyTypeEnum<'ctx>,
+    ) -> Result<(), String> {
+        check_int_llvm_type(scrutinee, T::get_int_type(ctx))
+    }
+}
+
+impl<'ctx, T: IsFixedInt> Model<'ctx> for FixedIntModel<T> {
+    type Value = FixedInt<'ctx, T>;
+
+    fn get_llvm_type(&self, ctx: &'ctx Context) -> BasicTypeEnum<'ctx> {
+        T::get_int_type(ctx).as_basic_type_enum()
+    }
+
+    fn review(&self, ctx: &'ctx Context, value: AnyValueEnum<'ctx>) -> Self::Value {
+        let value = review_int_llvm_value(value, T::get_int_type(ctx)).unwrap();
+        FixedInt { int: self.0, value }
+    }
+}
+
+impl<T: IsFixedInt> FixedIntModel<T> {
+    pub fn to_int_model(self, ctx: &Context) -> IntModel<'_> {
+        IntModel(T::get_int_type(ctx))
+    }
+}
+
+/// An inhabitant of [`FixedIntModel<'ctx>`]
+#[derive(Debug, Clone, Copy)]
+pub struct FixedInt<'ctx, T: IsFixedInt> {
+    pub int: T,
+    pub value: IntValue<'ctx>,
+}
+
+// FixedInt's Implementation
+
+impl<'ctx, T: IsFixedInt> ModelValue<'ctx> for FixedInt<'ctx, T> {
+    fn get_llvm_value(&self) -> BasicValueEnum<'ctx> {
+        self.value.as_basic_value_enum()
+    }
+}
+
+impl<'ctx, T: IsFixedInt> FixedInt<'ctx, T> {
+    pub fn to_int(self) -> Int<'ctx> {
+        Int(self.value)
+    }
+
+    pub fn signed_cast_to_fixed<R: IsFixedInt>(
+        self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        target_fixed_int: R,
+        name: &str,
+    ) -> FixedInt<'ctx, R> {
+        FixedInt {
+            int: target_fixed_int,
+            value: ctx
+                .builder
+                .build_int_s_extend_or_bit_cast(self.value, R::get_int_type(ctx.ctx), name)
+                .unwrap(),
+        }
+    }
+}
--- a/nac3core/src/codegen/model/function_builder.rs
+++ b/nac3core/src/codegen/model/function_builder.rs
@ -0,0 +1,55 @@
+use inkwell::{
+    types::{BasicMetadataTypeEnum, BasicType},
+    values::{AnyValue, BasicMetadataValueEnum},
+};
+
+use crate::codegen::{model::*, CodeGenContext};
+
+// TODO: Variadic argument?
+pub struct FunctionBuilder<'ctx, 'a> {
+    ctx: &'a CodeGenContext<'ctx, 'a>,
+    fn_name: &'a str,
+    arguments: Vec<(BasicMetadataTypeEnum<'ctx>, BasicMetadataValueEnum<'ctx>)>,
+}
+
+impl<'ctx, 'a> FunctionBuilder<'ctx, 'a> {
+    pub fn begin(ctx: &'a CodeGenContext<'ctx, 'a>, fn_name: &'a str) -> Self {
+        FunctionBuilder { ctx, fn_name, arguments: Vec::new() }
+    }
+
+    // NOTE: `_name` is for self-documentation
+    #[must_use]
+    pub fn arg<M: Model<'ctx>>(mut self, _name: &'static str, model: M, value: M::Value) -> Self {
+        model.check(self.ctx.ctx, value); // Panics if the passed `value` has the incorrect type.
+
+        self.arguments
+            .push((model.get_llvm_type(self.ctx.ctx).into(), value.get_llvm_value().into()));
+        self
+    }
+
+    pub fn returning<M: Model<'ctx>>(self, name: &'static str, return_model: M) -> M::Value {
+        let (param_tys, param_vals): (Vec<_>, Vec<_>) = self.arguments.into_iter().unzip();
+
+        let function = self.ctx.module.get_function(self.fn_name).unwrap_or_else(|| {
+            let return_type = return_model.get_llvm_type(self.ctx.ctx);
+            let fn_type = return_type.fn_type(&param_tys, false);
+            self.ctx.module.add_function(self.fn_name, fn_type, None)
+        });
+
+        let ret = self.ctx.builder.build_call(function, &param_vals, name).unwrap();
+        return_model.review(self.ctx.ctx, ret.as_any_value_enum())
+    }
+
+    // TODO: Code duplication, but otherwise returning<S: Optic<'ctx>> cannot resolve S if return_optic = None
+    pub fn returning_void(self) {
+        let (param_tys, param_vals): (Vec<_>, Vec<_>) = self.arguments.into_iter().unzip();
+
+        let function = self.ctx.module.get_function(self.fn_name).unwrap_or_else(|| {
+            let return_type = self.ctx.ctx.void_type();
+            let fn_type = return_type.fn_type(&param_tys, false);
+            self.ctx.module.add_function(self.fn_name, fn_type, None)
+        });
+
+        self.ctx.builder.build_call(function, &param_vals, "").unwrap();
+    }
+}
--- a/nac3core/src/codegen/model/int.rs
+++ b/nac3core/src/codegen/model/int.rs
@ -0,0 +1,97 @@
+use inkwell::{
+    context::Context,
+    types::{AnyType, AnyTypeEnum, BasicType, BasicTypeEnum, IntType},
+    values::{AnyValueEnum, BasicValue, BasicValueEnum, IntValue},
+};
+
+use crate::codegen::CodeGenContext;
+
+use super::{core::*, int_util::check_int_llvm_type, FixedInt, IsFixedInt};
+
+/// A model representing an [`IntType<'ctx>`].
+///
+/// Also see [`FixedIntModel`], which is more constrained than [`IntModel`]
+/// but provides more type-safe mechanisms and even auto-derivation of [`BasicTypeEnum<'ctx>`]
+/// for creating LLVM structures.
+#[derive(Debug, Clone, Copy)]
+pub struct IntModel<'ctx>(pub IntType<'ctx>);
+
+impl<'ctx> CanCheckLLVMType<'ctx> for IntModel<'ctx> {
+    fn check_llvm_type(
+        &self,
+        _ctx: &'ctx Context,
+        scrutinee: AnyTypeEnum<'ctx>,
+    ) -> Result<(), String> {
+        check_int_llvm_type(scrutinee, self.0)
+    }
+}
+
+impl<'ctx> Model<'ctx> for IntModel<'ctx> {
+    type Value = Int<'ctx>;
+
+    fn get_llvm_type(&self, _ctx: &'ctx Context) -> BasicTypeEnum<'ctx> {
+        self.0.as_basic_type_enum()
+    }
+
+    fn review(&self, ctx: &'ctx Context, value: AnyValueEnum<'ctx>) -> Self::Value {
+        let int = value.into_int_value();
+        self.check_llvm_type(ctx, int.get_type().as_any_type_enum()).unwrap();
+        Int(int)
+    }
+}
+
+impl<'ctx> IntModel<'ctx> {
+    /// Create a constant value that inhabits this [`IntModel<'ctx>`].
+    #[must_use]
+    pub fn constant(&self, value: u64) -> Int<'ctx> {
+        Int(self.0.const_int(value, false))
+    }
+
+    /// Check if `other` is fully compatible with this [`IntModel<'ctx>`].
+    ///
+    /// This simply checks if the underlying [`IntType<'ctx>`] has
+    /// the same number of bits.
+    #[must_use]
+    pub fn same_as(&self, other: IntModel<'ctx>) -> bool {
+        // TODO: or `self.0 == other.0` would also work?
+        self.0.get_bit_width() == other.0.get_bit_width()
+    }
+}
+
+/// An inhabitant of an [`IntModel<'ctx>`]
+#[derive(Debug, Clone, Copy)]
+pub struct Int<'ctx>(pub IntValue<'ctx>);
+
+impl<'ctx> ModelValue<'ctx> for Int<'ctx> {
+    fn get_llvm_value(&self) -> BasicValueEnum<'ctx> {
+        self.0.as_basic_value_enum()
+    }
+}
+
+impl<'ctx> Int<'ctx> {
+    #[must_use]
+    pub fn signed_cast_to_int(
+        self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        target_int: IntModel<'ctx>,
+        name: &str,
+    ) -> Int<'ctx> {
+        Int(ctx.builder.build_int_s_extend_or_bit_cast(self.0, target_int.0, name).unwrap())
+    }
+
+    #[must_use]
+    pub fn signed_cast_to_fixed<T: IsFixedInt>(
+        self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        target_fixed: T,
+        name: &str,
+    ) -> FixedInt<'ctx, T> {
+        FixedInt {
+            int: target_fixed,
+            value: ctx
+                .builder
+                .build_int_s_extend_or_bit_cast(self.0, T::get_int_type(ctx.ctx), name)
+                .unwrap(),
+        }
+    }
+}
--- a/nac3core/src/codegen/model/int_util.rs
+++ b/nac3core/src/codegen/model/int_util.rs
@ -0,0 +1,39 @@
+use inkwell::{
+    types::{AnyType, AnyTypeEnum, IntType},
+    values::{AnyValueEnum, IntValue},
+};
+
+/// Helper function to check if `scrutinee` is the same as `expected_int_type`
+pub fn check_int_llvm_type<'ctx>(
+    scrutinee: AnyTypeEnum<'ctx>,
+    expected_int_type: IntType<'ctx>,
+) -> Result<(), String> {
+    // Check if llvm_type is int type
+    let AnyTypeEnum::IntType(scrutinee) = scrutinee else {
+        return Err(format!("Expecting an int type but got {scrutinee:?}"));
+    };
+
+    // Check bit width
+    if scrutinee.get_bit_width() != expected_int_type.get_bit_width() {
+        return Err(format!(
+            "Expecting an int type of {}-bit(s) but got int type {}-bit(s)",
+            expected_int_type.get_bit_width(),
+            scrutinee.get_bit_width()
+        ));
+    }
+
+    Ok(())
+}
+
+/// Helper function to cast `scrutinee` is into an [`IntValue<'ctx>`].
+/// The LLVM type of `scrutinee` will be checked with [`check_int_llvm_type`].
+pub fn review_int_llvm_value<'ctx>(
+    value: AnyValueEnum<'ctx>,
+    expected_int_type: IntType<'ctx>,
+) -> Result<IntValue<'ctx>, String> {
+    // Check if value is of int type, error if that is anything else
+    check_int_llvm_type(value.get_type().as_any_type_enum(), expected_int_type)?;
+
+    // Ok, it is must be an int
+    Ok(value.into_int_value())
+}
--- a/nac3core/src/codegen/model/mod.rs
+++ b/nac3core/src/codegen/model/mod.rs
@ -0,0 +1,18 @@
+pub mod core;
+pub mod fixed_int;
+pub mod function_builder;
+pub mod int;
+mod int_util;
+pub mod opaque;
+pub mod pointer;
+pub mod slice;
+pub mod structure;
+
+pub use core::*;
+pub use fixed_int::*;
+pub use function_builder::*;
+pub use int::*;
+pub use opaque::*;
+pub use pointer::*;
+pub use slice::*;
+pub use structure::*;
--- a/nac3core/src/codegen/model/opaque.rs
+++ b/nac3core/src/codegen/model/opaque.rs
@ -0,0 +1,46 @@
+use inkwell::{
+    context::Context,
+    types::{AnyTypeEnum, BasicTypeEnum},
+    values::{AnyValueEnum, BasicValueEnum},
+};
+
+use super::*;
+
+#[derive(Debug, Clone, Copy)]
+pub struct OpaqueModel<'ctx>(pub BasicTypeEnum<'ctx>);
+
+impl<'ctx> CanCheckLLVMType<'ctx> for OpaqueModel<'ctx> {
+    fn check_llvm_type(
+        &self,
+        _ctx: &'ctx Context,
+        scrutinee: AnyTypeEnum<'ctx>,
+    ) -> Result<(), String> {
+        match BasicTypeEnum::try_from(scrutinee) {
+            Ok(_) => Ok(()),
+            Err(_err) => Err(format!("Expecting a BasicTypeEnum, but got {scrutinee:?}")),
+        }
+    }
+}
+
+impl<'ctx> Model<'ctx> for OpaqueModel<'ctx> {
+    type Value = Opaque<'ctx>;
+
+    fn get_llvm_type(&self, _ctx: &'ctx Context) -> BasicTypeEnum<'ctx> {
+        self.0
+    }
+
+    fn review(&self, ctx: &'ctx Context, value: AnyValueEnum<'ctx>) -> Self::Value {
+        self.check_llvm_type(ctx, value.get_type()).unwrap();
+        let value = BasicValueEnum::try_from(value).unwrap(); // Must work
+        Opaque(value)
+    }
+}
+
+#[derive(Debug, Clone, Copy)]
+pub struct Opaque<'ctx>(pub BasicValueEnum<'ctx>);
+
+impl<'ctx> ModelValue<'ctx> for Opaque<'ctx> {
+    fn get_llvm_value(&self) -> BasicValueEnum<'ctx> {
+        self.0
+    }
+}
--- a/nac3core/src/codegen/model/pointer.rs
+++ b/nac3core/src/codegen/model/pointer.rs
@ -0,0 +1,83 @@
+use inkwell::{
+    context::Context,
+    types::{AnyTypeEnum, BasicType, BasicTypeEnum},
+    values::{AnyValue, AnyValueEnum, BasicValue, BasicValueEnum, PointerValue},
+    AddressSpace,
+};
+
+use crate::codegen::CodeGenContext;
+
+use super::{core::*, OpaqueModel};
+
+/// A [`Model<'ctx>`] representing an LLVM [`PointerType<'ctx>`]
+/// with *full* information on the element u
+///
+/// [`self.0`] contains [`Model<'ctx>`] that represents the
+/// LLVM type of element of the [`PointerType<'ctx>`] is pointing at
+/// (like `PointerType<'ctx>::get_element_type()`, but abstracted as a [`Model<'ctx>`]).
+#[derive(Debug, Clone, Copy, Default)]
+pub struct PointerModel<E>(pub E);
+
+impl<'ctx, E: Model<'ctx>> CanCheckLLVMType<'ctx> for PointerModel<E> {
+    fn check_llvm_type(
+        &self,
+        ctx: &'ctx Context,
+        scrutinee: AnyTypeEnum<'ctx>,
+    ) -> Result<(), String> {
+        // Check if scrutinee is even a PointerValue
+        let AnyTypeEnum::PointerType(scrutinee) = scrutinee else {
+            return Err(format!("Expecting a pointer value, but got {scrutinee:?}"));
+        };
+
+        // Check the type of what the pointer is pointing at
+        // TODO: This will be deprecated by inkwell > llvm14 because `get_element_type()` will be gone
+        self.0.check_llvm_type(ctx, scrutinee.get_element_type())?; // TODO: Include backtrace?
+
+        Ok(())
+    }
+}
+
+impl<'ctx, E: Model<'ctx>> Model<'ctx> for PointerModel<E> {
+    type Value = Pointer<'ctx, E>;
+
+    fn get_llvm_type(&self, ctx: &'ctx Context) -> BasicTypeEnum<'ctx> {
+        self.0.get_llvm_type(ctx).ptr_type(AddressSpace::default()).as_basic_type_enum()
+    }
+
+    fn review(&self, ctx: &'ctx Context, value: AnyValueEnum<'ctx>) -> Self::Value {
+        self.check_llvm_type(ctx, value.get_type()).unwrap();
+
+        // TODO: Check get_element_type(). For inkwell LLVM 14 at least...
+        Pointer { element: self.0, value: value.into_pointer_value() }
+    }
+}
+
+/// An inhabitant of [`PointerModel<E>`]
+#[derive(Debug, Clone, Copy)]
+pub struct Pointer<'ctx, E: Model<'ctx>> {
+    pub element: E,
+    pub value: PointerValue<'ctx>,
+}
+
+impl<'ctx, E: Model<'ctx>> ModelValue<'ctx> for Pointer<'ctx, E> {
+    fn get_llvm_value(&self) -> BasicValueEnum<'ctx> {
+        self.value.as_basic_value_enum()
+    }
+}
+
+impl<'ctx, E: Model<'ctx>> Pointer<'ctx, E> {
+    /// Build an instruction to store a value into this pointer
+    pub fn store(&self, ctx: &CodeGenContext<'ctx, '_>, val: E::Value) {
+        ctx.builder.build_store(self.value, val.get_llvm_value()).unwrap();
+    }
+
+    /// Build an instruction to load a value from this pointer
+    pub fn load(&self, ctx: &CodeGenContext<'ctx, '_>, name: &str) -> E::Value {
+        let val = ctx.builder.build_load(self.value, name).unwrap();
+        self.element.review(ctx.ctx, val.as_any_value_enum())
+    }
+
+    pub fn to_opaque(self, ctx: &'ctx Context) -> Pointer<'ctx, OpaqueModel<'ctx>> {
+        Pointer { element: OpaqueModel(self.element.get_llvm_type(ctx)), value: self.value }
+    }
+}
--- a/nac3core/src/codegen/model/slice.rs
+++ b/nac3core/src/codegen/model/slice.rs
@ -0,0 +1,73 @@
+use crate::codegen::{CodeGenContext, CodeGenerator};
+
+use super::{Int, Model, Pointer};
+
+pub struct ArraySlice<'ctx, E: Model<'ctx>> {
+    pub num_elements: Int<'ctx>,
+    pub pointer: Pointer<'ctx, E>,
+}
+
+impl<'ctx, E: Model<'ctx>> ArraySlice<'ctx, E> {
+    pub fn ix_unchecked(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        idx: Int<'ctx>,
+        name: &str,
+    ) -> Pointer<'ctx, E> {
+        let element_addr =
+            unsafe { ctx.builder.build_in_bounds_gep(self.pointer.value, &[idx.0], name).unwrap() };
+        Pointer { value: element_addr, element: self.pointer.element }
+    }
+
+    pub fn ix<G: CodeGenerator + ?Sized>(
+        &self,
+        generator: &mut G,
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        idx: Int<'ctx>,
+        name: &str,
+    ) -> Pointer<'ctx, E> {
+        let int_type = self.num_elements.0.get_type(); // NOTE: Weird get_type(), see comment under `trait Ixed`
+
+        assert_eq!(int_type.get_bit_width(), idx.0.get_type().get_bit_width()); // Might as well check bit width to catch bugs
+
+        // TODO: SGE or UGE? or make it defined by the implementee?
+
+        // Check `0 <= index`
+        let lower_bounded = ctx
+            .builder
+            .build_int_compare(
+                inkwell::IntPredicate::SLE,
+                int_type.const_zero(),
+                idx.0,
+                "lower_bounded",
+            )
+            .unwrap();
+
+        // Check `index < num_elements`
+        let upper_bounded = ctx
+            .builder
+            .build_int_compare(
+                inkwell::IntPredicate::SLT,
+                idx.0,
+                self.num_elements.0,
+                "upper_bounded",
+            )
+            .unwrap();
+
+        // Compute `0 <= index && index < num_elements`
+        let bounded = ctx.builder.build_and(lower_bounded, upper_bounded, "bounded").unwrap();
+
+        // Assert `bounded`
+        ctx.make_assert(
+            generator,
+            bounded,
+            "0:IndexError",
+            "nac3core LLVM codegen attempting to access out of bounds array index {0}. Must satisfy 0 <= index < {2}",
+            [ Some(idx.0), Some(self.num_elements.0), None],
+            ctx.current_loc
+        );
+
+        // ...and finally do indexing
+        self.ix_unchecked(ctx, idx, name)
+    }
+}
--- a/nac3core/src/codegen/model/structure.rs
+++ b/nac3core/src/codegen/model/structure.rs
@ -0,0 +1,219 @@
+use inkwell::{
+    context::Context,
+    types::{AnyType, AnyTypeEnum, BasicType, BasicTypeEnum, StructType},
+    values::{AnyValueEnum, BasicValue, BasicValueEnum, StructValue},
+};
+use itertools::{izip, Itertools};
+
+use crate::codegen::CodeGenContext;
+
+use super::{core::CanCheckLLVMType, Model, ModelValue, Pointer};
+
+#[derive(Debug, Clone, Copy)]
+pub struct Field<E> {
+    pub gep_index: u64,
+    pub name: &'static str,
+    pub element: E,
+}
+
+struct FieldLLVM<'ctx> {
+    gep_index: u64,
+    name: &'ctx str,
+    llvm_type: BasicTypeEnum<'ctx>,
+
+    // Only CanCheckLLVMType is needed, dont use `Model<'ctx>`
+    llvm_type_model: Box<dyn CanCheckLLVMType<'ctx> + 'ctx>,
+}
+
+pub struct FieldBuilder<'ctx> {
+    pub ctx: &'ctx Context,
+    gep_index_counter: u64,
+    struct_name: &'ctx str,
+    fields: Vec<FieldLLVM<'ctx>>,
+}
+
+impl<'ctx> FieldBuilder<'ctx> {
+    #[must_use]
+    pub fn new(ctx: &'ctx Context, struct_name: &'ctx str) -> Self {
+        FieldBuilder { ctx, gep_index_counter: 0, struct_name, fields: Vec::new() }
+    }
+
+    fn next_gep_index(&mut self) -> u64 {
+        let index = self.gep_index_counter;
+        self.gep_index_counter += 1;
+        index
+    }
+
+    pub fn add_field<E: Model<'ctx> + 'ctx>(&mut self, name: &'static str, element: E) -> Field<E> {
+        let gep_index = self.next_gep_index();
+
+        self.fields.push(FieldLLVM {
+            gep_index,
+            name,
+            llvm_type: element.get_llvm_type(self.ctx),
+            llvm_type_model: Box::new(element),
+        });
+
+        Field { gep_index, name, element }
+    }
+
+    pub fn add_field_auto<E: Model<'ctx> + Default + 'ctx>(
+        &mut self,
+        name: &'static str,
+    ) -> Field<E> {
+        self.add_field(name, E::default())
+    }
+}
+
+/// A marker trait to mark singleton struct that describes a particular LLVM structure.
+pub trait IsStruct<'ctx>: Clone + Copy {
+    /// The type of the Rust `struct` that holds all the fields of this LLVM struct.
+    type Fields;
+
+    /// A cosmetic name for this struct.
+    /// TODO: Currently unused. To be used in error reporting.
+    fn struct_name(&self) -> &'static str;
+
+    fn build_fields(&self, builder: &mut FieldBuilder<'ctx>) -> Self::Fields;
+
+    fn get_fields(&self, ctx: &'ctx Context) -> Self::Fields {
+        let mut builder = FieldBuilder::new(ctx, self.struct_name());
+        self.build_fields(&mut builder)
+    }
+
+    /// Get the LLVM struct type this [`IsStruct<'ctx>`] is representing.
+    fn get_struct_type(&self, ctx: &'ctx Context) -> StructType<'ctx> {
+        let mut builder = FieldBuilder::new(ctx, self.struct_name());
+        self.build_fields(&mut builder); // Self::Fields is discarded
+
+        let field_types = builder.fields.iter().map(|f| f.llvm_type).collect_vec();
+        ctx.struct_type(&field_types, false)
+    }
+
+    /// Check if `scrutinee` matches the [`StructType<'ctx>`] this [`IsStruct<'ctx>`] is representing.
+    fn check_struct_type(
+        &self,
+        ctx: &'ctx Context,
+        scrutinee: StructType<'ctx>,
+    ) -> Result<(), String> {
+        // Details about scrutinee
+        let scrutinee_field_types = scrutinee.get_field_types();
+
+        // Details about the defined specifications of this struct
+        // We will access them through builder
+        let mut builder = FieldBuilder::new(ctx, self.struct_name());
+        self.build_fields(&mut builder);
+
+        // Check # of fields
+        if builder.fields.len() != scrutinee_field_types.len() {
+            return Err(format!(
+                "Expecting struct to have {} field(s), but scrutinee has {} field(s)",
+                builder.fields.len(),
+                scrutinee_field_types.len()
+            ));
+        }
+
+        // Check the types of each field
+        // TODO: Traceback?
+        for (f, scrutinee_field_type) in izip!(builder.fields, scrutinee_field_types) {
+            f.llvm_type_model.check_llvm_type(ctx, scrutinee_field_type.as_any_type_enum())?;
+        }
+
+        Ok(())
+    }
+}
+
+/// A [`Model<'ctx>`] that represents an LLVM struct.
+///
+/// `self.0` contains a [`IsStruct<'ctx>`] that gives the details of the LLVM struct.
+#[derive(Debug, Clone, Copy, Default)]
+pub struct StructModel<S>(pub S);
+
+impl<'ctx, S: IsStruct<'ctx>> CanCheckLLVMType<'ctx> for StructModel<S> {
+    fn check_llvm_type(
+        &self,
+        ctx: &'ctx Context,
+        scrutinee: AnyTypeEnum<'ctx>,
+    ) -> Result<(), String> {
+        // Check if scrutinee is even a struct type
+        let AnyTypeEnum::StructType(scrutinee) = scrutinee else {
+            return Err(format!("Expecting a struct type, but got {scrutinee:?}"));
+        };
+
+        // Ok. now check the struct type *thoroughly*
+        self.0.check_struct_type(ctx, scrutinee)
+    }
+}
+
+impl<'ctx, S: IsStruct<'ctx>> Model<'ctx> for StructModel<S> {
+    type Value = Struct<'ctx, S>;
+
+    fn get_llvm_type(&self, ctx: &'ctx Context) -> BasicTypeEnum<'ctx> {
+        self.0.get_struct_type(ctx).as_basic_type_enum()
+    }
+
+    fn review(&self, ctx: &'ctx Context, value: AnyValueEnum<'ctx>) -> Self::Value {
+        // Check that `value` is not some bogus values or an incorrect StructValue
+        self.check_llvm_type(ctx, value.get_type()).unwrap();
+
+        Struct { structure: self.0, value: value.into_struct_value() }
+    }
+}
+
+#[derive(Debug, Clone, Copy)]
+pub struct Struct<'ctx, S> {
+    pub structure: S,
+    pub value: StructValue<'ctx>,
+}
+
+impl<'ctx, S: IsStruct<'ctx>> ModelValue<'ctx> for Struct<'ctx, S> {
+    fn get_llvm_value(&self) -> BasicValueEnum<'ctx> {
+        self.value.as_basic_value_enum()
+    }
+}
+
+impl<'ctx, S: IsStruct<'ctx>> Pointer<'ctx, StructModel<S>> {
+    /// Build an instruction that does `getelementptr` on an LLVM structure referenced by this pointer.
+    ///
+    /// This provides a nice syntax to chain up `getelementptr` in an intuitive and type-safe way:
+    ///
+    /// ```ignore
+    /// let ctx: &CodeGenContext<'ctx, '_>;
+    /// let ndarray: Pointer<'ctx, StructModel<NpArray<'ctx>>>;
+    /// ndarray.gep(ctx, |f| f.ndims).store();
+    /// ```
+    ///
+    /// You might even write chains `gep`, i.e.,
+    /// ```ignore
+    /// my_struct
+    ///     .gep(ctx, |f| f.thing1)
+    ///     .gep(ctx, |f| f.value)
+    ///     .store(ctx, my_value) // Equivalent to `my_struct.thing1.value = my_value`
+    /// ```
+    pub fn gep<E, GetFieldFn>(
+        &self,
+        ctx: &CodeGenContext<'ctx, '_>,
+        get_field: GetFieldFn,
+    ) -> Pointer<'ctx, E>
+    where
+        E: Model<'ctx>,
+        GetFieldFn: FnOnce(S::Fields) -> Field<E>,
+    {
+        let fields = self.element.0.get_fields(ctx.ctx);
+        let field = get_field(fields);
+
+        let llvm_i32 = ctx.ctx.i32_type(); // TODO: I think I'm not supposed to *just* use i32 for GEP like that
+
+        let ptr = unsafe {
+            ctx.builder
+                .build_in_bounds_gep(
+                    self.value,
+                    &[llvm_i32.const_zero(), llvm_i32.const_int(field.gep_index, false)],
+                    field.name,
+                )
+                .unwrap()
+        };
+
+        Pointer { element: field.element, value: ptr }
+    }
+}
--- a/nac3core/src/codegen/numpy.rs
+++ b/nac3core/src/codegen/numpy.rs
@ -9,7 +9,7 @@ use crate::{
        irrt::{
            calculate_len_for_slice_range, call_ndarray_calc_broadcast,
            call_ndarray_calc_broadcast_index, call_ndarray_calc_nd_indices,
-            call_ndarray_calc_size,
+            call_ndarray_calc_size, error_context::call_nac3_dummy_raise,
        },
        llvm_intrinsics::{self, call_memcpy_generic},
        stmt::{gen_for_callback_incrementing, gen_for_range_callback, gen_if_else_expr_callback},
@ -1742,6 +1742,8 @@ pub fn gen_ndarray_zeros<'ctx>(
    assert!(obj.is_none());
    assert_eq!(args.len(), 1);

+    call_nac3_dummy_raise(generator, context);
+
    let shape_ty = fun.0.args[0].ty;
    let shape_arg = args[0].1.clone().to_basic_value_enum(context, generator, shape_ty)?;

--- a/nac3core/src/lib.rs
+++ b/nac3core/src/lib.rs
@ -23,3 +23,4 @@ pub mod codegen;
 pub mod symbol_resolver;
 pub mod toplevel;
 pub mod typecheck;
+pub(crate) mod util;
--- a/nac3core/src/toplevel/builtins.rs
+++ b/nac3core/src/toplevel/builtins.rs
@ -23,6 +23,7 @@ use crate::{
    symbol_resolver::SymbolValue,
    toplevel::{helper::PrimDef, numpy::make_ndarray_ty},
    typecheck::typedef::{into_var_map, iter_type_vars, TypeVar, VarMap},
+    util::SizeVariant,
 };

 use super::*;
@ -278,20 +279,11 @@ pub fn get_builtins(unifier: &mut Unifier, primitives: &PrimitiveStore) -> Built
        .collect()
 }

-/// A helper enum used by [`BuiltinBuilder`]
-#[derive(Clone, Copy)]
-enum SizeVariant {
-    Bits32,
-    Bits64,
-}
-
-impl SizeVariant {
-    fn of_int(self, primitives: &PrimitiveStore) -> Type {
-        match self {
+fn get_size_variant_of_int(size_variant: SizeVariant, primitives: &PrimitiveStore) -> Type {
+    match size_variant {
        SizeVariant::Bits32 => primitives.int32,
        SizeVariant::Bits64 => primitives.int64,
    }
-    }
 }

 struct BuiltinBuilder<'a> {
@ -1061,7 +1053,7 @@ impl<'a> BuiltinBuilder<'a> {
        );

        // The size variant of the function determines the size of the returned int.
-        let int_sized = size_variant.of_int(self.primitives);
+        let int_sized = get_size_variant_of_int(size_variant, self.primitives);

        let ndarray_int_sized =
            make_ndarray_ty(self.unifier, self.primitives, Some(int_sized), Some(common_ndim.ty));
@ -1086,7 +1078,7 @@ impl<'a> BuiltinBuilder<'a> {
                let arg_ty = fun.0.args[0].ty;
                let arg = args[0].1.clone().to_basic_value_enum(ctx, generator, arg_ty)?;

-                let ret_elem_ty = size_variant.of_int(&ctx.primitives);
+                let ret_elem_ty = get_size_variant_of_int(size_variant, &ctx.primitives);
                Ok(Some(builtin_fns::call_round(generator, ctx, (arg_ty, arg), ret_elem_ty)?))
            }),
        )
@ -1127,7 +1119,7 @@ impl<'a> BuiltinBuilder<'a> {
            make_ndarray_ty(self.unifier, self.primitives, Some(float), Some(common_ndim.ty));

        // The size variant of the function determines the type of int returned
-        let int_sized = size_variant.of_int(self.primitives);
+        let int_sized = get_size_variant_of_int(size_variant, self.primitives);
        let ndarray_int_sized =
            make_ndarray_ty(self.unifier, self.primitives, Some(int_sized), Some(common_ndim.ty));

@ -1150,7 +1142,7 @@ impl<'a> BuiltinBuilder<'a> {
                let arg_ty = fun.0.args[0].ty;
                let arg = args[0].1.clone().to_basic_value_enum(ctx, generator, arg_ty)?;

-                let ret_elem_ty = size_variant.of_int(&ctx.primitives);
+                let ret_elem_ty = get_size_variant_of_int(size_variant, &ctx.primitives);
                let func = match kind {
                    Kind::Ceil => builtin_fns::call_ceil,
                    Kind::Floor => builtin_fns::call_floor,
--- a/nac3core/src/util.rs
+++ b/nac3core/src/util.rs
@ -0,0 +1,6 @@
+/// A helper enum used by [`BuiltinBuilder`]
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum SizeVariant {
+    Bits32,
+    Bits64,
+}
Author	SHA1	Message	Date
lyken	eac164ce11	WIP	2024-07-18 16:07:47 +08:00
lyken	e8aa6129f0	core/irrt: add c-string utils	2024-07-17 17:50:55 +08:00
lyken	e1abb819c6	core: refactor to lift `struct SizeVariant` to /util.rs	2024-07-17 17:50:49 +08:00
lyken	db9e9586b5	core/model: add var_alloc & array_var_alloc	2024-07-17 17:50:39 +08:00
lyken	2525369760	core/model: add FunctionBuilder	2024-07-17 17:50:31 +08:00
lyken	06b64ea888	core/model: introduce `Model<'ctx>` abstraction	2024-07-17 17:50:23 +08:00
lyken	37a022e156	core/irrt: introduce irrt testing `cargo test -F test` would compile `nac3core/irrt/irrt_test.cpp` targetted to the host machine (it gets to use `std`) and run the test executable.	2024-07-17 12:14:17 +08:00
lyken	17e8eeaa46	core/irrt: split irrt.cpp into headers & reformat To prepare organizing for future IRRT implementations. There will be a lot of C++ code.	2024-07-17 12:14:15 +08:00
lyken	eb3aa18ce1	core/irrt: build.rs capture IR defined constants	2024-07-17 12:14:13 +08:00
lyken	8068bd5cb0	core/irrt: build.rs capture IR defined types	2024-07-17 12:14:08 +08:00
lyken	3a8c4009fd	core/irrt: comment build.rs & move irrt to its own dir To prepare for future IRRT implementations, and to also make cargo only have to watch a single directory.	2024-07-17 12:14:05 +08:00
lyken	ca13f56404	core/irrt: build with -W=return-type To help catch future C++ programming bugs	2024-07-17 12:14:02 +08:00