stage1 更多的语义以及类型构造器

.gitignore

@@ -8,5 +8,6 @@ build/
 *.o
 *.obj
 
+*.old
 *.exe
 *.out

build.py

@@ -1,417 +0,0 @@
-#!/usr/bin/env python3
-"""SPL build system - .c -> .o -> exe, .spl -> .sir, deps auto-resolved."""
-
-import argparse
-import hashlib
-import json
-import os
-import shutil
-import subprocess
-import sys
-from pathlib import Path
-
-ROOT = Path(__file__).resolve().parent
-BUILD_DIR = ROOT / "build"
-CACHE_FILE = BUILD_DIR / ".build_cache.json"
-DESC_FILE = ROOT / "project_desc.py"
-EXE_EXT = ".exe" if os.name == "nt" else ""
-
-CC = os.environ.get("CC", "gcc")
-CFLAGS = os.environ.get("CFLAGS", "-Wall -Wextra -O0 -g").split()
-
-def load_desc():
-    ns = {"__builtins__": __builtins__}
-    exec(DESC_FILE.read_text(), ns)
-    return {
-        "exe": ns.get("exe", {}),
-        "spl": ns.get("spl", {}),
-        "pipeline": ns.get("pipeline", {}),
-    }
-
-# ── Helpers ─────────────────────────────────────────────────────────────
-
-def obj_key(src):
-    return src.rsplit(".", 1)[0].replace("/", "_").replace("\\", "_")
-
-def obj_path(src):
-    return BUILD_DIR / (obj_key(src) + ".o")
-
-def exe_path(name):
-    return BUILD_DIR / (name + EXE_EXT)
-
-def sir_path(name):
-    return BUILD_DIR / (name + ".sir")
-
-# ── Cache ───────────────────────────────────────────────────────────────
-
-def load_cache():
-    if CACHE_FILE.is_file():
-        try:
-            return json.loads(CACHE_FILE.read_text())
-        except (json.JSONDecodeError, OSError):
-            pass
-    return {}
-
-def save_cache(cache):
-    BUILD_DIR.mkdir(parents=True, exist_ok=True)
-    CACHE_FILE.write_text(json.dumps(cache, indent=2))
-
-def hash_file(p):
-    h = hashlib.sha256()
-    with p.open("rb") as f:
-        while chunk := f.read(65536):
-            h.update(chunk)
-    return h.hexdigest()
-
-# ── C compilation ───────────────────────────────────────────────────────
-
-_built_objs = set()
-
-def compile_c(src, cache, force):
-    key = f"obj:{src}"
-    dst = obj_path(src)
-    src_path = ROOT / src
-
-    if not src_path.is_file():
-        print(f"  ERR  {src} not found"); return False
-
-    h = hashlib.sha256()
-    h.update(hash_file(src_path).encode())
-    h.update(" ".join(CFLAGS).encode())
-    ch = h.hexdigest()
-
-    cached = cache.get(key)
-    if not force and cached and cached["h"] == ch and dst.is_file():
-        _built_objs.add(src); return True
-
-    BUILD_DIR.mkdir(parents=True, exist_ok=True)
-    cmd = [CC] + CFLAGS + ["-c", str(src_path), "-o", str(dst)]
-    print(f"  CC    {src}")
-    if subprocess.run(cmd).returncode: return False
-    cache[key] = {"h": ch}
-    _built_objs.add(src)
-    return True
-
-def link_exe(name, sources, cache, force):
-    key = f"exe:{name}"
-    dst = exe_path(name)
-
-    # hash = concat of all input obj hashes + name
-    h = hashlib.sha256(name.encode())
-    for src in sources:
-        cached = cache.get(f"obj:{src}", {})
-        h.update(cached.get("h", "?").encode())
-    ch = h.hexdigest()
-
-    cached = cache.get(key)
-    if not force and cached and cached["h"] == ch and dst.is_file():
-        return True
-
-    cmd = [CC] + CFLAGS + [str(obj_path(s)) for s in sources] + ["-o", str(dst)]
-    print(f"  LINK  {name}")
-    if subprocess.run(cmd).returncode: return False
-    cache[key] = {"h": ch}
-    return True
-
-# ── SPL compilation ─────────────────────────────────────────────────────
-
-def find_imports(src_path):
-    """Scan source for @import("path") directives, return resolved paths."""
-    deps = []
-    try:
-        text = src_path.read_text(errors="replace")
-    except OSError:
-        return deps
-    idx = 0
-    while True:
-        idx = text.find('@import("', idx)
-        if idx < 0:
-            break
-        start = idx + 9
-        end = text.find('")', start)
-        if end < 0:
-            break
-        imp_path = text[start:end]
-        imp_file = (src_path.parent / imp_path).resolve()
-        if imp_file.is_file():
-            deps.append(imp_file)
-        idx = end + 2
-    return deps
-
-def collect_deps(src_path):
-    """Collect all transitive @import dependencies."""
-    seen = set()
-    result = []
-    def walk(p):
-        p = p.resolve()
-        if p in seen:
-            return
-        seen.add(p)
-        for dep in find_imports(p):
-            walk(dep)
-            result.append(dep)
-    walk(src_path)
-    return result
-
-def compile_spl(name, src, compiler, desc, cache, force):
-    key = f"spl:{name}"
-    dst = sir_path(name)
-    src_path = ROOT / src
-
-    if not src_path.is_file():
-        print(f"  ERR  {src} not found"); return False
-
-    h = hashlib.sha256()
-    h.update(hash_file(src_path).encode())
-
-    # Hash @import dependencies (transitive)
-    for dep in collect_deps(src_path):
-        h.update(hash_file(dep).encode())
-
-    # Determine how to run the compiler
-    if compiler in desc.get("spl", {}):
-        # SPL-based compiler: run via VM runner
-        cinfo = desc["spl"][compiler]
-        pl = desc.get("pipeline", {}).get(cinfo.get("pipeline", ""), {})
-        runner = pl.get("runner") or cinfo.get("runner", "spl_cli")
-        runner_exe = exe_path(runner)
-        sir = sir_path(compiler)
-        if runner_exe.is_file():
-            h.update(hash_file(runner_exe).encode())
-        if os.path.isfile(sir):
-            h.update(hash_file(Path(sir)).encode())
-        ch = h.hexdigest()
-        cached = cache.get(key)
-        if not force and cached and cached["h"] == ch and dst.is_file():
-            return True
-        BUILD_DIR.mkdir(parents=True, exist_ok=True)
-        cmd = [str(runner_exe), str(sir), str(src_path), str(dst)]
-        print(f"  SPL   {name}  ({compiler} via {runner})")
-    else:
-        # Native exe compiler
-        ce = exe_path(compiler)
-        if ce.is_file():
-            h.update(hash_file(ce).encode())
-        ch = h.hexdigest()
-        cached = cache.get(key)
-        if not force and cached and cached["h"] == ch and dst.is_file():
-            return True
-        BUILD_DIR.mkdir(parents=True, exist_ok=True)
-        cmd = [str(ce), str(src_path), str(dst)]
-        print(f"  SPL   {name}  ({compiler})")
-
-    if subprocess.run(cmd).returncode: return False
-    cache[key] = {"h": ch}
-    return True
-
-# ── Dependency-aware build ──────────────────────────────────────────────
-
-_built = set()
-
-def build(name, desc, cache, force):
-    """Build target + all deps (auto-resolved DAG traversal)."""
-    if name in _built:
-        return True
-
-    if name in desc.get("exe", {}):
-        sources = desc["exe"][name]
-        for src in sources:
-            if not compile_c(src, cache, force):
-                return False
-        if not link_exe(name, sources, cache, force):
-            return False
-
-    elif name in desc.get("spl", {}):
-        info = desc["spl"][name]
-        pl = desc.get("pipeline", {}).get(info.get("pipeline", ""), {})
-        compiler = pl.get("compiler") or info.get("compiler")
-        if not compiler:
-            print(f"  ERR  '{name}': no compiler in pipeline '{info.get('pipeline')}'")
-            return False
-        if not build(compiler, desc, cache, force):
-            return False
-        if not compile_spl(name, info["src"], compiler, desc, cache, force):
-            return False
-
-    else:
-        print(f"  ERR  unknown target '{name}'")
-        return False
-
-    _built.add(name)
-    return True
-
-def build_all(desc, cache, force):
-    targets = list(desc.get("exe", {})) + list(desc.get("spl", {}))
-    for t in targets:
-        if not build(t, desc, cache, force):
-            return False
-    return True
-
-# ── Run ─────────────────────────────────────────────────────────────────
-
-def run_target(name, desc, cache, force, extra_args):
-    if name in desc.get("exe", {}):
-        if not build(name, desc, cache, force):
-            return 1
-        exe = exe_path(name)
-        print(f"  RUN   {name}")
-        return subprocess.run([str(exe)] + extra_args).returncode
-
-    if name in desc.get("spl", {}):
-        info = desc["spl"][name]
-        pl = desc.get("pipeline", {}).get(info.get("pipeline", ""), {})
-        runner = pl.get("runner") or info.get("runner", "spl_cli")
-        if not runner:
-            print(f"  ERR  '{name}': no runner"); return 1
-        if not build(name, desc, cache, force):
-            return 1
-        if not build(runner, desc, cache, force):
-            return 1
-        print(f"  RUN   {name}  (via {runner})")
-        return subprocess.run(
-            [str(exe_path(runner)), str(sir_path(name))] + extra_args
-        ).returncode
-
-    if name in desc.get("pipeline", {}):
-        pl = desc["pipeline"][name]
-        compiler = pl.get("compiler")
-        runner = pl.get("runner")
-        if not compiler or not runner:
-            print(f"  ERR  pipeline '{name}' missing compiler or runner"); return 1
-        if not extra_args:
-            print("  ERR  usage: run <pipeline> <src> [args]"); return 1
-        src = extra_args[0]
-
-        if not build(compiler, desc, cache, force): return 1
-        if not build(runner, desc, cache, force): return 1
-
-        sir_name = "pipeline_" + hashlib.sha256(str(ROOT / src).encode()).hexdigest()[:8]
-        if not compile_spl(sir_name, src, compiler, desc, cache, force):
-            return 1
-
-        print(f"  RUN   pipeline {name}  ({src})")
-        return subprocess.run(
-            [str(exe_path(runner)), str(sir_path(sir_name))] + extra_args[1:]
-        ).returncode
-
-    print(f"  ERR  unknown target '{name}'"); return 1
-
-# ── Clean / List ────────────────────────────────────────────────────────
-
-def clean():
-    if BUILD_DIR.is_dir():
-        shutil.rmtree(BUILD_DIR)
-        print(f"  CLEAN {BUILD_DIR}")
-
-def list_targets(desc):
-    print("Executables (.c -> .o -> exe):")
-    for name, sources in desc.get("exe", {}).items():
-        print(f"  {name}")
-        for s in sources:
-            print(f"    {s}")
-    if desc.get("spl"):
-        print("\nSPL programs (.spl -> .sir via pipeline):")
-        for name, info in desc["spl"].items():
-            pl_name = info.get("pipeline", "-")
-            print(f"  {name}  pipeline={pl_name}  src={info['src']}")
-    if desc.get("pipeline"):
-        print("\nPipelines:")
-        for name, pl in desc["pipeline"].items():
-            print(f"  {name}  compiler={pl['compiler']}  runner={pl.get('runner', '-')}")
-
-# ── CLI ─────────────────────────────────────────────────────────────────
-
-def main():
-    parser = argparse.ArgumentParser(description="SPL build system")
-    sub = parser.add_subparsers(dest="command")
-
-    bp = sub.add_parser("build", help="build targets (default: all)")
-    bp.add_argument("targets", nargs="*")
-    bp.add_argument("-f", "--force", action="store_true")
-
-    rp = sub.add_parser("run", help="build and run a target, or run pipeline <name> <src> [args]")
-    rp.add_argument("target")
-    rp.add_argument("extra", nargs="*")
-
-    sub.add_parser("test", help="build and run VM unit tests")
-    sub.add_parser("clean", help="remove build artifacts")
-    sub.add_parser("list", help="list all targets")
-
-    args = parser.parse_args()
-
-    if not DESC_FILE.is_file():
-        print(f"ERR  {DESC_FILE} not found"); return 1
-
-    # Default: build all
-    if args.command is None:
-        desc = load_desc(); cache = load_cache()
-        ok = build_all(desc, cache, False)
-        save_cache(cache)
-        return 0 if ok else 1
-
-    # Build
-    if args.command == "build":
-        desc = load_desc(); cache = load_cache()
-        if args.targets:
-            ok = all(build(t, desc, cache, args.force) for t in args.targets)
-        else:
-            ok = build_all(desc, cache, args.force)
-        save_cache(cache)
-        return 0 if ok else 1
-
-    # Run
-    if args.command == "run":
-        desc = load_desc(); cache = load_cache()
-        rc = run_target(args.target, desc, cache, False, args.extra)
-        save_cache(cache)
-        return rc
-
-    # Pipeline: build compiler -> compile .spl -> run via runner
-    if args.command == "pipeline":
-        desc = load_desc(); cache = load_cache()
-        pl = desc.get("pipeline", {}).get(args.name)
-        if not pl:
-            print(f"  ERR  unknown pipeline '{args.name}'"); return 1
-        compiler = pl.get("compiler")
-        runner = pl.get("runner")
-        if not compiler or not runner:
-            print(f"  ERR  pipeline '{args.name}' missing compiler or runner"); return 1
-        src_path = ROOT / args.src
-        if not src_path.is_file():
-            print(f"  ERR  source not found: {args.src}"); return 1
-
-        # Build compiler + runner
-        if not build(compiler, desc, cache, False): return 1
-        if not build(runner, desc, cache, False): return 1
-
-        # Compile .spl -> .sir (temp name = src path hash)
-        sir_name = "pipeline_" + hashlib.sha256(str(src_path).encode()).hexdigest()[:8]
-        if not compile_spl(sir_name, args.src, compiler, desc, cache, False):
-            return 1
-
-        # Run .sir via runner
-        print(f"  PIPELINE {args.name}  ({args.src})")
-        rc = subprocess.run(
-            [str(exe_path(runner)), str(sir_path(sir_name))] + args.extra
-        ).returncode
-        save_cache(cache)
-        return rc
-
-    # Test
-    if args.command == "test":
-        desc = load_desc(); cache = load_cache()
-        ok = build("test", desc, cache, False)
-        save_cache(cache)
-        if not ok: return 1
-        return subprocess.run([str(exe_path("test"))]).returncode
-
-    # Clean / List
-    if args.command == "clean":
-        clean(); return 0
-    if args.command == "list":
-        list_targets(load_desc()); return 0
-
-    return 0
-
-if __name__ == "__main__":
-    sys.exit(main())

nob.c

@@ -0,0 +1,294 @@
+#undef UNICODE
+#define NOB_IMPLEMENTATION
+// Redefine nob_cc_flags() to carry the project's debug flags. Keep in sync
+// with CFLAGS_STR below.
+#define SPL_CFLAGS "-Wall", "-Wextra", "-O0", "-g", "-D_CRT_SECURE_NO_WARNINGS"
+#define nob_cc_flags(cmd) nob_cmd_append(cmd, SPL_CFLAGS)
+#include "nob.h"
+#ifdef _WIN32
+#include <consoleapi2.h>
+#endif
+
+// C-side bootstrap build script. Builds the stage0 VM tools and the splc0
+// compiler from their C sources. The SPL self-hosting chain (splc1, ...) is
+// out of scope and is driven by a different mechanism.
+//
+// Usage:
+//     cc -o nob nob.c          # bootstrap once
+//     ./nob                    # build all targets
+//     ./nob build <targets...> # build specific targets
+//     ./nob run <target> [..]  # build and run
+//     ./nob test               # build and run VM unit tests
+//     ./nob clean              # remove build artifacts
+//     ./nob list               # list all targets
+//     ./nob help               # print this help
+
+#define BUILD_FOLDER "build/"
+
+#ifdef _WIN32
+#define EXE_SUFFIX ".exe"
+#else
+#define EXE_SUFFIX ""
+#endif
+
+// Content of the .cflags stamp file; must match SPL_CFLAGS above.
+#define CFLAGS_STR "-Wall -Wextra -O0 -g"
+// Stamp file whose mtime is refreshed whenever the flags change so that all
+// object files get rebuilt.
+static const char *CFLAGS_STAMP = BUILD_FOLDER ".cflags";
+
+// Target descriptions
+#define VM_SRCS "stage0/spl_mcode.c", "stage0/spl_syscall.c", "stage0/spl_vm.c"
+#define SPLC0_PART_SRCS                                                                            \
+    "stage1/spl_ast.c", "stage1/spl_lexer.c", "stage1/spl_dumptree.c", "stage1/spl_type.c",        \
+        "stage1/spl_sema.c", "stage1/spl_ir.c", "stage1/spl_ast2ir.c"
+
+static const char *spl_cli_srcs[] = {"stage0/spl_cli.c", VM_SRCS};
+static const char *splc0_srcs[] = {"stage1/splc0.c", VM_SRCS, SPLC0_PART_SRCS};
+static const char *test_srcs[] = {"stage0/test_spl_vm.c", VM_SRCS};
+static const char *spl_disasm_srcs[] = {"stage0/spl_disasm.c", VM_SRCS};
+
+typedef struct {
+    const char *name;
+    const char **srcs;
+    size_t srcs_count;
+} Exe_Target;
+
+static Exe_Target exe_targets[] = {
+    {"spl_cli", spl_cli_srcs, NOB_ARRAY_LEN(spl_cli_srcs)},
+    {"splc0", splc0_srcs, NOB_ARRAY_LEN(splc0_srcs)},
+    {"test", test_srcs, NOB_ARRAY_LEN(test_srcs)},
+    {"spl_disasm", spl_disasm_srcs, NOB_ARRAY_LEN(spl_disasm_srcs)},
+};
+
+static const Exe_Target *find_exe_target(const char *name) {
+    for (size_t i = 0; i < NOB_ARRAY_LEN(exe_targets); ++i) {
+        if (strcmp(exe_targets[i].name, name) == 0) {
+            return &exe_targets[i];
+        }
+    }
+    return NULL;
+}
+
+static const char *obj_path(const char *src) {
+    // "stage0/spl_vm.c" -> "build/stage0_spl_vm.o"
+    size_t prefix_len = strlen(BUILD_FOLDER);
+    char *out = (char *)nob_temp_alloc(prefix_len + strlen(src) + 2 + 1);
+    char *w = out;
+    memcpy(w, BUILD_FOLDER, prefix_len);
+    w += prefix_len;
+    for (const char *p = src; *p && *p != '.'; ++p) {
+        *w++ = (*p == '/' || *p == '\\') ? '_' : *p;
+    }
+    memcpy(w, ".o", 3);
+    return out;
+}
+
+static const char *exe_path(const char *name) {
+    return nob_temp_sprintf(BUILD_FOLDER "%s" EXE_SUFFIX, name);
+}
+
+static bool sync_cflags_stamp(void) {
+    Nob_String_View cflags = nob_sv_from_cstr(CFLAGS_STR);
+    Nob_String_Builder old = {0};
+    bool changed = true;
+    if (nob_file_exists(CFLAGS_STAMP) && nob_read_entire_file(CFLAGS_STAMP, &old)) {
+        changed = !nob_sv_eq(nob_sb_to_sv(old), cflags);
+    }
+    nob_sb_free(old);
+
+    if (changed) {
+        if (!nob_write_entire_file(CFLAGS_STAMP, cflags.data, cflags.count)) {
+            return false;
+        }
+    }
+    return true;
+}
+
+static bool prepare_build(void) {
+    if (!nob_mkdir_if_not_exists(BUILD_FOLDER))
+        return false;
+    return sync_cflags_stamp();
+}
+
+static bool build_object(const char *src) {
+    const char *obj = obj_path(src);
+    if (!nob_file_exists(src)) {
+        nob_log(NOB_ERROR, "source file `%s` does not exist", src);
+        return false;
+    }
+
+    const char *inputs[] = {src, CFLAGS_STAMP};
+    int needs = nob_needs_rebuild(obj, inputs, NOB_ARRAY_LEN(inputs));
+    if (needs < 0)
+        return false;
+    if (needs == 0)
+        return true;
+
+    Nob_Cmd cmd = {0};
+    nob_cc(&cmd);
+    nob_cc_flags(&cmd);
+    nob_cmd_append(&cmd, "-c");
+    nob_cc_inputs(&cmd, src);
+    nob_cc_output(&cmd, obj);
+    return nob_cmd_run(&cmd);
+}
+
+static bool link_exe(const Exe_Target *target) {
+    const char *exe = exe_path(target->name);
+    const char **objs = (const char **)nob_temp_alloc(sizeof(const char *) * target->srcs_count);
+    for (size_t i = 0; i < target->srcs_count; ++i) {
+        objs[i] = obj_path(target->srcs[i]);
+    }
+
+    int needs = nob_needs_rebuild(exe, objs, target->srcs_count);
+    if (needs < 0)
+        return false;
+    if (needs == 0)
+        return true;
+
+    Nob_Cmd cmd = {0};
+    nob_cc(&cmd);
+    nob_cc_flags(&cmd);
+    for (size_t i = 0; i < target->srcs_count; ++i) {
+        nob_cc_inputs(&cmd, objs[i]);
+    }
+    nob_cc_output(&cmd, exe);
+    return nob_cmd_run(&cmd);
+}
+
+static bool build_target(const char *name) {
+    const Exe_Target *target = find_exe_target(name);
+    if (target == NULL) {
+        nob_log(NOB_ERROR, "unknown target `%s`", name);
+        return false;
+    }
+
+    for (size_t i = 0; i < target->srcs_count; ++i) {
+        if (!build_object(target->srcs[i])) {
+            return false;
+        }
+    }
+    return link_exe(target);
+}
+
+static bool build_all(void) {
+    for (size_t i = 0; i < NOB_ARRAY_LEN(exe_targets); ++i) {
+        if (!build_target(exe_targets[i].name)) {
+            return false;
+        }
+    }
+    return true;
+}
+
+static bool delete_walk_entry(Nob_Walk_Entry entry) { return nob_delete_file(entry.path); }
+
+static bool delete_directory_recursively(const char *dir_path) {
+    return nob_walk_dir(dir_path, delete_walk_entry, .post_order = true);
+}
+
+static void print_help(const char *program_name) {
+    nob_log(NOB_INFO, "SPL C-side bootstrap build system (nob)");
+    nob_log(NOB_INFO, "%s", " ");
+    nob_log(NOB_INFO, "Usage: %s <command> [args...]", program_name);
+    nob_log(NOB_INFO, "%s", " ");
+    nob_log(NOB_INFO, "Commands:");
+    nob_log(NOB_INFO, "    build [targets...]  build all targets or the given ones");
+    nob_log(NOB_INFO, "    run <target> [args] build and run a target");
+    nob_log(NOB_INFO, "    test                build and run VM unit tests");
+    nob_log(NOB_INFO, "    clean               remove build artifacts");
+    nob_log(NOB_INFO, "    list                list all targets");
+    nob_log(NOB_INFO, "    help/-h/--help      print this help message");
+    nob_log(NOB_INFO, "%s", " ");
+    nob_log(NOB_INFO, "Targets:");
+    for (size_t i = 0; i < NOB_ARRAY_LEN(exe_targets); ++i) {
+        nob_log(NOB_INFO, "    %s", exe_targets[i].name);
+    }
+}
+
+int main(int argc, char **argv) {
+#ifdef _WIN32
+    SetConsoleOutputCP(CP_UTF8);
+#endif
+    NOB_GO_REBUILD_URSELF_PLUS(argc, argv, "nob.h");
+    set_log_handler(nob_cancer_log_handler);
+
+    const char *program_name = nob_shift(argv, argc);
+    const char *command_name = "build";
+    if (argc > 0) {
+        command_name = nob_shift(argv, argc);
+    }
+
+    if (strcmp(command_name, "build") == 0) {
+        if (!prepare_build())
+            return 1;
+        if (argc > 0) {
+            while (argc > 0) {
+                const char *target_name = nob_shift(argv, argc);
+                if (!build_target(target_name))
+                    return 1;
+            }
+        } else {
+            if (!build_all())
+                return 1;
+        }
+        return 0;
+    }
+
+    if (strcmp(command_name, "run") == 0) {
+        if (argc <= 0) {
+            nob_log(NOB_ERROR, "usage: %s run <target> [args...]", program_name);
+            return 1;
+        }
+        const char *target_name = nob_shift(argv, argc);
+        if (!prepare_build())
+            return 1;
+        if (!build_target(target_name))
+            return 1;
+        Nob_Cmd cmd = {0};
+        nob_cmd_append(&cmd, exe_path(target_name));
+        while (argc > 0) {
+            nob_cmd_append(&cmd, nob_shift(argv, argc));
+        }
+        return nob_cmd_run(&cmd) ? 0 : 1;
+    }
+
+    if (strcmp(command_name, "test") == 0) {
+        if (!prepare_build())
+            return 1;
+        if (!build_target("test"))
+            return 1;
+        Nob_Cmd cmd = {0};
+        nob_cmd_append(&cmd, exe_path("test"));
+        return nob_cmd_run(&cmd) ? 0 : 1;
+    }
+
+    if (strcmp(command_name, "clean") == 0) {
+        if (nob_file_exists(BUILD_FOLDER)) {
+            if (!delete_directory_recursively(BUILD_FOLDER))
+                return 1;
+        }
+        nob_log(NOB_INFO, "cleaned %s", BUILD_FOLDER);
+        return 0;
+    }
+
+    if (strcmp(command_name, "list") == 0) {
+        for (size_t i = 0; i < NOB_ARRAY_LEN(exe_targets); ++i) {
+            nob_log(NOB_INFO, "%s:", exe_targets[i].name);
+            for (size_t j = 0; j < exe_targets[i].srcs_count; ++j) {
+                nob_log(NOB_INFO, "    %s", exe_targets[i].srcs[j]);
+            }
+        }
+        return 0;
+    }
+
+    if (strcmp(command_name, "help") == 0 || strcmp(command_name, "-h") == 0 ||
+        strcmp(command_name, "--help") == 0) {
+        print_help(program_name);
+        return 0;
+    }
+
+    nob_log(NOB_ERROR, "unknown command `%s`", command_name);
+    print_help(program_name);
+    return 1;
+}

project_desc.py

@@ -1,53 +0,0 @@
-vm = [
-    "stage0/spl_mcode.c",
-    "stage0/spl_syscall.c",
-    "stage0/spl_vm.c",
-]
-
-splc0_part = [
-    # "stage1/spl_ir.c",
-    "stage1/spl_ast.c",
-    "stage1/spl_lexer.c",
-    "stage1/spl_dumptree.c",
-    # "stage1/spl_type.c",
-    # "stage1/spl_sema.c",
-    # "stage1/spl_builtin.c",
-    # "stage1/spl_ast2ir.c",
-    # "stage1/spl_ir2vm.c",
-]
-
-exe = {
-    "spl_cli":   ["stage0/spl_cli.c"] + vm,
-    "splc0":     ["stage1/splc0.c"] + vm + splc0_part,
-    "test":      ["stage0/test_spl_vm.c"] + vm,
-    "spl_disasm": ["stage0/spl_disasm.c"] + vm,
-}
-
-pipeline = {
-    "splc0b": {"compiler": "splc0", "runner": "spl_cli"},
-    "splc0r": {"compiler": "splc0", "runner": "spl_cli"},
-    "splc0d": {"compiler": "splc0", "runner": "spl_disasm"},
-
-    "splc1b": {"compiler": "splc0", "runner": "spl_cli"},
-    "splc1r": {"compiler": "splc1", "runner": "spl_cli"},
-    "splc1d": {"compiler": "splc1", "runner": "spl_disasm"},
-
-    "splc2b": {"compiler": "splc1", "runner": "spl_cli"},
-    "splc2r": {"compiler": "splc2", "runner": "spl_cli"},
-    "splc2d": {"compiler": "splc2", "runner": "spl_disasm"},
-
-    "splc3b": {"compiler": "splc2", "runner": "spl_cli"},
-    "splc3r": {"compiler": "splc3", "runner": "spl_cli"},
-    "splc3d": {"compiler": "splc3", "runner": "spl_disasm"},
-
-    "splc4b": {"compiler": "splc3", "runner": "spl_cli"},
-    "splc4r": {"compiler": "splc4", "runner": "spl_cli"},
-    "splc4d": {"compiler": "splc4", "runner": "spl_disasm"},
-}
-
-spl = {
-    "splc1": {"src": "stage2/splc1.spl", "pipeline": "splc0r"},
-    "splc2": {"src": "stage2/splc2.spl", "pipeline": "splc1r"},
-    "splc3": {"src": "stage3/splc3.spl", "pipeline": "splc2r"},
-    "splc4": {"src": "stage4/splc4.spl", "pipeline": "splc3r"},
-}

stage0/include/core_map.h

@@ -77,6 +77,8 @@ static inline usize map_hash_str(const char *s) {
     for (usize(idx) = 0; (idx) < (map).cap; ++(idx))                                               \
         if ((map).data[(idx)].state == __MAP_SLOT_OCCUPIED)
 
+#define unsafe_map_at(map, idx) ((map).data[(idx)])
+
 /**
  * 插入（若键已存在则更新值）
  * 注意：扩容使用 realloc，失败会 abort（可自行修改错误处理）
@@ -128,6 +130,56 @@ static inline usize map_hash_str(const char *s) {
         }                                                                                          \
     } while (0)
 
+/**
+ * 插入（仅当键不存在时），若键已存在则静默跳过，不更新值
+ * 与 map_put 的差异仅在"键已存在时"：map_put 更新值，map_put_new 放弃
+ */
+#define map_put_new(map, _key, _val)                                                               \
+    do {                                                                                           \
+        /* 扩容 */                                                                                 \
+        if ((map).cap == 0 || (map).size * 128 / (map).cap >= MAP_DEFAULT_LOAD_FACTOR) {           \
+            usize new_cap = (map).cap == 0 ? 8 : (map).cap * 2;                                    \
+            MAP_SLOT(MAP_TYPEOF((map).data->key), MAP_TYPEOF((map).data->val)) *new_data =         \
+                calloc(new_cap, sizeof(*new_data));                                                \
+            if (!new_data)                                                                         \
+                abort();                                                                           \
+            /* 重新插入旧元素 */                                                                   \
+            for (usize _i = 0; _i < (map).cap; ++_i) {                                             \
+                if ((map).data[_i].state == __MAP_SLOT_OCCUPIED) {                                 \
+                    usize _h = (map).hash((map).data[_i].key) & (new_cap - 1);                     \
+                    while (new_data[_h].state == __MAP_SLOT_OCCUPIED)                              \
+                        _h = (_h + 1) & (new_cap - 1);                                             \
+                    new_data[_h].key = (map).data[_i].key;                                         \
+                    new_data[_h].val = (map).data[_i].val;                                         \
+                    new_data[_h].state = __MAP_SLOT_OCCUPIED;                                      \
+                }                                                                                  \
+            }                                                                                      \
+            free((map).data);                                                                      \
+            (map).data = (void *)new_data;                                                         \
+            (map).cap = new_cap;                                                                   \
+        }                                                                                          \
+        /* 查找 */                                                                                 \
+        usize _mask = (map).cap - 1;                                                               \
+        usize _idx = (map).hash(_key) & _mask;                                                     \
+        usize _first_del = (usize) - 1;                                                            \
+        while ((map).data[_idx].state != __MAP_SLOT_EMPTY) {                                       \
+            if ((map).data[_idx].state == __MAP_SLOT_OCCUPIED &&                                   \
+                (map).cmp((map).data[_idx].key, _key) == 0) {                                      \
+                break; /* 已存在，不更新 */                                                         \
+            }                                                                                      \
+            if ((map).data[_idx].state == __MAP_SLOT_DELETED && _first_del == (usize) - 1)         \
+                _first_del = _idx;                                                                 \
+            _idx = (_idx + 1) & _mask;                                                             \
+        }                                                                                          \
+        if ((map).data[_idx].state == __MAP_SLOT_EMPTY) {                                          \
+            usize _target = (_first_del != (usize) - 1) ? _first_del : _idx;                       \
+            (map).data[_target].key = _key;                                                        \
+            (map).data[_target].val = _val;                                                        \
+            (map).data[_target].state = __MAP_SLOT_OCCUPIED;                                       \
+            ++(map).size;                                                                          \
+        }                                                                                          \
+    } while (0)
+
 /**
  * 查询：若找到，*out_val 被赋值为对应值并返回 1；否则返回 0
  */
@@ -150,6 +202,33 @@ static inline usize map_hash_str(const char *s) {
         _found;                                                                                    \
     }))
 
+/**
+ * 查询（续探）：首次调用前将 *in_out_idx 置为 (usize)-1，
+ * 则从键自身的 hash 桶开始查找；命中后 *out_val 被赋值、返回 1，
+ * 且 *in_out_idx 更新为命中槽位，下次调用从该槽的下一个继续探测（碰撞续探）。
+ * 扫描到空槽返回 0。用于 key 碰撞后需要回表二次比较的场景。
+ */
+#define map_get_continue(map, _key, out_val, in_out_idx)                                           \
+    (({                                                                                            \
+        int _found = 0;                                                                            \
+        if ((map).cap > 0) {                                                                       \
+            usize _mask = (map).cap - 1;                                                           \
+            usize _idx = (*(in_out_idx) == (usize) - 1) ? (map).hash(_key) & _mask                 \
+                                                        : (*(in_out_idx) + 1) & _mask;            \
+            while ((map).data[_idx].state != __MAP_SLOT_EMPTY) {                                   \
+                if ((map).data[_idx].state == __MAP_SLOT_OCCUPIED &&                               \
+                    (map).cmp((map).data[_idx].key, _key) == 0) {                                  \
+                    *out_val = (map).data[_idx].val;                                               \
+                    *(in_out_idx) = _idx;                                                          \
+                    _found = 1;                                                                    \
+                    break;                                                                         \
+                }                                                                                  \
+                _idx = (_idx + 1) & _mask;                                                         \
+            }                                                                                      \
+        }                                                                                          \
+        _found;                                                                                    \
+    }))
+
 /**
  * 删除指定键
  */

stage0/spl_disasm.c

@@ -54,19 +54,19 @@ int main(int argc, const char **argv) {
         spl_vm_ins_t *ins = &vec_at(prog.insns, i);
         printf("%4zd: %s", i, spl_vm_opcode_name((spl_vm_opcode_t)ins->opcode));
         if (ins->type != SPL_VOID)
-            printf(" %s", spl_vm_type_kind_name((spl_type_t)ins->type));
+            printf(" %s", spl_vm_type_kind_name((spl_vm_kind_t)ins->type));
         printf(" %zd", ins->imm);
 
         /* annotate jumps / calls */
         if (ins->opcode == SPL_JMP || ins->opcode == SPL_BZ || ins->opcode == SPL_BNZ) {
-            long long target = (long long)i + 1 + (long long)ins->imm;
-            printf("  ; -> %zd", target);
+            spl_vm_val_t target = i + 1 + ins->imm;
+            printf("  ; -> %zu", target);
         } else if (ins->opcode == SPL_CALL) {
-            printf("  ; nargs=%zd, from stack", (long long)ins->imm);
+            printf("  ; nargs=%zu, from stack", ins->imm);
         } else if (ins->opcode == SPL_CALLI) {
             printf("  ; indirect call");
         } else if (ins->opcode == SPL_GADDR) {
-            printf("  ; gdata[%zd]", (long long)ins->imm);
+            printf("  ; gdata[%zu]", ins->imm);
         }
         printf("\n");
     }

stage0/spl_mcode.c

@@ -427,7 +427,7 @@ const char *opcode_name[] = {
 #undef X
 };
 const char *spl_vm_opcode_name(spl_vm_opcode_t opcode) { return opcode_name[opcode]; }
-const char *spl_vm_type_kind_name(spl_type_t type) {
+const char *spl_vm_type_kind_name(spl_vm_kind_t type) {
     switch (type) {
     case SPL_VOID:
         return "void";
@@ -467,7 +467,7 @@ const char *spl_vm_type_kind_name(spl_type_t type) {
 void spl_vm_ins_dump(spl_vm_ins_t *ins, spl_vm_val_t addr) {
     printf("%4zu: %s", addr, spl_vm_opcode_name((spl_vm_opcode_t)ins->opcode));
     if (ins->type != SPL_VOID)
-        printf(" %s", spl_vm_type_kind_name((spl_type_t)ins->type));
+        printf(" %s", spl_vm_type_kind_name((spl_vm_kind_t)ins->type));
     printf(" %zd:%zx", ins->imm, ins->imm);
     printf("\n");
 }

stage0/spl_mcode.h

@@ -29,7 +29,7 @@ typedef enum {
     SPL_F64,
     SPL_PTR,
     SPL_TYPE_COUNT,
-} spl_type_t;
+} spl_vm_kind_t;
 
 /* clang-format off */
 #define SPL_OPCODES(X) \
@@ -213,7 +213,7 @@ spl_vm_native_t *spl_prog_get_native(spl_prog_t *prog, const char *name);
 
 /* Opcode name lookup for debugging/dumping */
 const char *spl_vm_opcode_name(spl_vm_opcode_t opcode);
-const char *spl_vm_type_kind_name(spl_type_t type);
+const char *spl_vm_type_kind_name(spl_vm_kind_t type);
 
 void spl_vm_ins_dump(spl_vm_ins_t *ins, spl_vm_val_t addr);
 

stage0/spl_vm.c

@@ -28,8 +28,8 @@
  * Type helpers
  * ================================================================ */
 
-static int spl_is_float(spl_type_t t) { return t == SPL_F32 || t == SPL_F64; }
-static int spl_is_signed(spl_type_t t) {
+static int spl_is_float(spl_vm_kind_t t) { return t == SPL_F32 || t == SPL_F64; }
+static int spl_is_signed(spl_vm_kind_t t) {
     switch (t) {
     case SPL_I8:
     case SPL_I16:
@@ -41,7 +41,7 @@ static int spl_is_signed(spl_type_t t) {
         return 0;
     }
 }
-static int spl_type_size(spl_type_t t) {
+static int spl_type_size(spl_vm_kind_t t) {
     switch (t) {
     case SPL_VOID:
         return 0;
@@ -123,7 +123,7 @@ static int spl_type_size(spl_type_t t) {
     do {                                                                                           \
         spl_vm_val_t _b = POP(), _a = POP();                                                       \
         spl_vm_val_t _r = 0;                                                                       \
-        if (spl_is_float((spl_type_t)ins->type)) {                                                 \
+        if (spl_is_float((spl_vm_kind_t)ins->type)) {                                              \
             double _da, _db, _dr;                                                                  \
             if (ins->type == SPL_F32) {                                                            \
                 float _fa, _fb;                                                                    \
@@ -185,7 +185,7 @@ static int spl_type_size(spl_type_t t) {
         if (_b == 0)                                                                               \
             VM_ERROR("division by zero");                                                          \
         spl_vm_val_t _r = 0;                                                                       \
-        if (spl_is_float((spl_type_t)ins->type)) {                                                 \
+        if (spl_is_float((spl_vm_kind_t)ins->type)) {                                              \
             double _da, _db, _dr;                                                                  \
             if (ins->type == SPL_F32) {                                                            \
                 float _fa, _fb;                                                                    \
@@ -545,6 +545,7 @@ LONG WINAPI UnhandledExceptionFilterImpl(EXCEPTION_POINTERS *pExceptionInfo) {
  * ================================================================ */
 void spl_vm_init_ex(spl_vm_t *vm, int stack_size, int call_depth) {
 #ifdef _WIN32
+    SetErrorMode(SEM_FAILCRITICALERRORS);
     SetUnhandledExceptionFilter(UnhandledExceptionFilterImpl);
     SetConsoleOutputCP(CP_UTF8);
     SetConsoleCP(CP_UTF8);
@@ -598,7 +599,10 @@ void spl_vm_add_breakpoint_fn(spl_vm_t *vm, const char *name) {
     for (usize i = 0; i < vec_size(vm->fn_breakpoints); i++)
         if (strcmp(vm->fn_breakpoints.data[i], name) == 0)
             return;
-    vec_push(vm->fn_breakpoints, strdup(name));
+    char *cpname = malloc(strlen(name) + 1);
+    Assert(cpname != NULL);
+    strcpy(cpname, name);
+    vec_push(vm->fn_breakpoints, cpname);
 }
 
 void spl_vm_clear_breakpoints(spl_vm_t *vm) {
@@ -843,7 +847,7 @@ int spl_vm_run_once(spl_vm_t *vm) {
 
     case SPL_NEG: {
         spl_vm_val_t _a = POP();
-        if (spl_is_float((spl_type_t)ins->type)) {
+        if (spl_is_float((spl_vm_kind_t)ins->type)) {
             double _d;
             if (ins->type == SPL_F32) {
                 float _f;

stage1/spl_ast.c

@@ -58,6 +58,7 @@ static inline spl_ast_node_t *node_at(spl_ast_t *ast, spl_ast_node_ref_t ref) {
 static spl_ast_node_ref_t new_node(parser_t *p, spl_ast_node_kind_t kind, const spl_tok_t *tok) {
     spl_ast_node_t n = {0};
     n.kind = kind;
+    n.resolved_def_id = 0;
     if (tok) {
         n.dbg.dbg_name = spl_ast_kind_name(kind);
         n.dbg.fname = tok->fname;
@@ -408,7 +409,7 @@ static spl_ast_node_ref_t parse_type_decl(parser_t *p, spl_ast_node_ref_vec_t at
     if (t && (t->type == TOK_COMMA || t->type == TOK_SEMICOLON)) {
         advance(p);
     } else {
-        LOG_WARN("need comma or semicolon with member decl");
+        SPL_WARN(t, "need comma or semicolon with member decl");
     }
     return ref;
 }
@@ -433,7 +434,7 @@ static spl_ast_node_ref_t parse_member_decl(parser_t *p, spl_ast_node_ref_vec_t
     if (t && (t->type == TOK_COMMA || t->type == TOK_SEMICOLON)) {
         advance(p);
     } else {
-        LOG_WARN("need comma or semicolon with member decl");
+        SPL_WARN(t, "need comma or semicolon with member decl");
     }
     return ref;
 }
@@ -483,7 +484,7 @@ static spl_ast_node_ref_t parse_var_const_decl(parser_t *p, spl_ast_node_ref_vec
     if (t && t->type == TOK_SEMICOLON) {
         advance(p);
     } else {
-        LOG_WARN("need semicolon in decl var/const");
+        SPL_WARN(t, "need semicolon in decl var/const");
     }
     return ref;
 }
@@ -1792,8 +1793,8 @@ void spl_ast_prase(spl_ast_t *ast) {
         if (d)
             vec_push(members, d);
     }
-    spl_ast_node_ref_t root = new_node(&p, SPL_AST_CONTAINER_ITEM, begin);
-    node_at(ast, root)->container_item.members = members;
+    spl_ast_node_ref_t root = new_node(&p, SPL_AST_CONTAINER_MEMBERS, begin);
+    node_at(ast, root)->container_members = members;
     ast->root = root;
     ast->parsed = p.failed ? -1 : 1;
 }
@@ -1809,6 +1810,12 @@ void spl_ast_drop(spl_ast_t *ast) {
     ast->root = 0;
 }
 
+spl_ast_node_t *spl_ast_node(spl_ast_t *ast, spl_ast_node_ref_t node) {
+    if (!ast || node == 0)
+        return NULL;
+    return &vec_at(ast->node_buckets, node);
+}
+
 void spl_ast_valid(spl_ast_t *ast) {
     if (!ast)
         return;
@@ -1827,6 +1834,7 @@ void spl_ast_dump(spl_ast_t *ast, spl_ast_node_ref_t node) {
         return;
     dump_stack_t stack;
     vec_init(stack);
+    printf("AST:\n");
     dump_node(ast, node, &stack, 1);
 }
 
@@ -1840,9 +1848,10 @@ static void dump_node(spl_ast_t *ast, spl_ast_node_ref_t node_ref, dump_stack_t
 
     for (usize i = 0; i < vec_size(*stack); i++)
         printf("%s", vec_at(*stack, i) ? "  " : "| ");
-    printf("%s%s #%zu", last ? "`-" : "|-", spl_ast_kind_name(node->kind), node_ref);
+    printf("%s%s #%zu ID`%zu`", last ? "`-" : "|-", spl_ast_kind_name(node->kind), node_ref,
+           node->resolved_type_id);
     if (node->dbg.fname)
-        printf(" (%d:%d)", node->dbg.line, node->dbg.col);
+        printf(" [%d:%d]", node->dbg.line, node->dbg.col);
     printf("\n");
 
     vec_push(*stack, last);
@@ -1851,17 +1860,18 @@ static void dump_node(spl_ast_t *ast, spl_ast_node_ref_t node_ref, dump_stack_t
         UNREACHABLE();
         break;
 
-    case SPL_AST_CONTAINER_ITEM:
-        dump_vec(ast, node->container_item.attr_list, stack);
-        dump_vec(ast, node->container_item.members, stack);
+    case SPL_AST_CONTAINER_MEMBERS:
+        dump_vec(ast, node->container_members, stack);
         break;
 
     case SPL_AST_FN_DECL:
     case SPL_AST_FN_DEFINE:
         dump_vec(ast, node->fn_decl.attr_list, stack);
         dump_vec(ast, node->fn_decl.param_list, stack);
-        if (node->fn_decl.type_expr)
-            dump_node(ast, node->fn_decl.type_expr, stack, 1);
+        if (node->fn_decl.type_expr) {
+            last = node->kind == SPL_AST_FN_DECL ? 1 : 0;
+            dump_node(ast, node->fn_decl.type_expr, stack, last);
+        }
         dump_vec(ast, node->fn_decl.block, stack);
         break;
 
@@ -2104,6 +2114,7 @@ static void dump_node(spl_ast_t *ast, spl_ast_node_ref_t node_ref, dump_stack_t
     case SPL_AST_TYPE_STRUCT:
     case SPL_AST_TYPE_UNION:
     case SPL_AST_TYPE_ENUM:
+    case SPL_AST_TYPE_SHAPE:
         dump_vec(ast, node->type_expr.attr_list, stack);
         dump_vec(ast, node->type_expr.aggregate_list, stack);
         break;

stage1/spl_ast.h

@@ -9,7 +9,7 @@
 /* clang-format off */
 #define SPL_AST_KIND_TABLE            \
     X(SPL_AST_NONE,                 V0, none) \
-    X(SPL_AST_CONTAINER_ITEM,       V0, container_item) \
+    X(SPL_AST_CONTAINER_MEMBERS,    V0, container_members) \
     X(SPL_AST_FN_DECL,              V0, fn_decl) \
     X(SPL_AST_FN_DEFINE,            V0, fn_define) \
     X(SPL_AST_TYPE_DECL,            V0, type_decl) \
@@ -97,6 +97,7 @@
     X(SPL_AST_TYPE_SLICE,           V0, type_slice) \
     X(SPL_AST_TYPE_STRUCT,          V0, type_struct) \
     X(SPL_AST_TYPE_UNION,           V0, type_union) \
+    X(SPL_AST_TYPE_SHAPE,           V0, type_shape) \
     X(SPL_AST_TYPE_ENUM,            V0, type_enum) \
     X(SPL_AST_TYPE_VOID,            V0, type_void) \
     X(SPL_AST_TYPE_BOOL,            V0, type_bool) \
@@ -138,12 +139,12 @@ struct spl_ast_node {
     spl_ast_node_kind_t kind;
     spl_dbg_node_t dbg;
 
-    usize resolved_def_id;
     union {
-        struct {
-            spl_ast_node_ref_vec_t attr_list; /* attr_item */
-            spl_ast_node_ref_vec_t members;   /* container_decl 列表 */
-        } container_item;
+        usize resolved_def_id;
+        usize resolved_type_id;
+    };
+    union {
+        spl_ast_node_ref_vec_t container_members; /* container_decl */
 
         struct {
             const char *ident;
@@ -154,7 +155,7 @@ struct spl_ast_node {
             spl_ast_node_ref_vec_t attr_list; /* attr_item */
             const char *name;
             spl_ast_node_ref_vec_t param_list; /* param_decl */
-            spl_ast_node_ref_t type_expr;
+            spl_ast_node_ref_t type_expr; /* ret type expr */
             spl_ast_node_ref_vec_t block; /* 语句/尾表达式 */
         } fn_decl;
         struct {
@@ -314,6 +315,8 @@ typedef struct {
 void spl_ast_init(spl_ast_t *ast, const spl_tok_vec_t *tok_vec /*move*/);
 void spl_ast_drop(spl_ast_t *ast);
 
+spl_ast_node_t* spl_ast_node(spl_ast_t *ast, spl_ast_node_ref_t node);
+
 void spl_ast_prase(spl_ast_t *ast);
 void spl_ast_valid(spl_ast_t *ast);
 void spl_ast_dump(spl_ast_t *ast, spl_ast_node_ref_t node);

stage1/spl_ast2ir.h

@@ -4,21 +4,13 @@
 #include "spl_ir.h"
 #include "spl_sema.h"
 
-/* 全局 var/const 的 def → gdata 向量中的 value 节点索引 */
-typedef struct {
-    spl_def_id_t def_id;
-    usize gdata_idx;
-} spl_ast2ir_gref_t;
-
 typedef struct {
     const spl_sema_t *sema;
-    spl_ir_t ir;
-    VEC(char *) owned_names;          /* 本模块 malloc 的函数名，drop 时释放 */
-    VEC(spl_ast2ir_gref_t) gdata_ref; /* def_id → gdata value 节点索引 */
+    spl_ir_builder_t *ir;
     int err_count;
 } spl_ast2ir_t;
 
-void spl_ast2ir_init(spl_ast2ir_t *ast2ir, const spl_sema_t *sema);
+void spl_ast2ir_init(spl_ast2ir_t *ast2ir, spl_ir_builder_t *ir, const spl_sema_t *sema);
 void spl_ast2ir_drop(spl_ast2ir_t *ast2ir);
 
 void spl_ast2ir_run(spl_ast2ir_t *ast2ir);

stage1/spl_dbg.h

@@ -1,8 +1,6 @@
 #ifndef __SPL_DBG_H__
 #define __SPL_DBG_H__
 
-#include "spl_tok.h"
-
 typedef struct {
     const char *fname;
     int line;
@@ -12,8 +10,16 @@ typedef struct {
     // TODO
 } spl_dbg_node_t;
 
-#define SPL_FATAL(tok, fmt, ...)                                                                   \
-    LOG_FATAL("fatal at %s:%zd:%zd " fmt, ((tok) && (tok)->fname ? (tok)->fname : "<null>"),       \
-              ((tok) ? (tok)->line : 0), ((tok) ? (tok)->col : 0), ##__VA_ARGS__)
+#define SPL_FATAL(dbg, fmt, ...)                                                                   \
+    LOG_FATAL("fatal at %s:%zd:%zd " fmt, ((dbg) && (dbg)->fname ? (dbg)->fname : "<null>"),       \
+              ((dbg) ? (dbg)->line : 0), ((dbg) ? (dbg)->col : 0), ##__VA_ARGS__)
+
+#define SPL_ERROR(dbg, fmt, ...)                                                                   \
+    LOG_ERROR("error at %s:%zd:%zd " fmt, ((dbg) && (dbg)->fname ? (dbg)->fname : "<null>"),       \
+              ((dbg) ? (dbg)->line : 0), ((dbg) ? (dbg)->col : 0), ##__VA_ARGS__)
+
+#define SPL_WARN(dbg, fmt, ...)                                                                    \
+    LOG_ERROR("warn at %s:%zd:%zd " fmt, ((dbg) && (dbg)->fname ? (dbg)->fname : "<null>"),        \
+              ((dbg) ? (dbg)->line : 0), ((dbg) ? (dbg)->col : 0), ##__VA_ARGS__)
 
 #endif /* __SPL_DBG_H__ */

stage1/spl_ir.h

@@ -5,7 +5,6 @@
 #include "spl_dbg.h"
 #include "spl_type.h"
 
-
 /* clang-format off */
 #define SPL_IR_FN_TABLE                                      \
     X(arith.add,           V0, SPL_IR_ARITH_ADD)             \
@@ -46,7 +45,6 @@
     X(mem.field,           V0, SPL_IR_MEM_FIELD_PTR)         \
     X(mem.copy,            V0, SPL_IR_MEM_COPY)              \
     X(mem.set,             V0, SPL_IR_MEM_SET)               \
-    X(mem.fence,           V0, SPL_IR_MEM_FENCE)             \
     X(type.const,          V0, SPL_IR_TYPE_CONST)            \
     X(type.bitsizeof,      V0, SPL_IR_TYPE_BITSIZEOF)        \
     X(type.sizeof,         V0, SPL_IR_TYPE_SIZEOF)           \
@@ -56,15 +54,6 @@
     X(agg.construct,       V0, SPL_IR_AGG_CONSTRUCT)         \
     X(agg.extract,         V0, SPL_IR_AGG_EXTRACT)           \
     X(agg.insert,          V0, SPL_IR_AGG_INSERT)            \
-    X(atomic.load,         V0, SPL_IR_ATOMIC_LOAD)           \
-    X(atomic.store,        V0, SPL_IR_ATOMIC_STORE)          \
-    X(atomic.rmw_add,      V0, SPL_IR_ATOMIC_RMW_ADD)        \
-    X(atomic.rmw_sub,      V0, SPL_IR_ATOMIC_RMW_SUB)        \
-    X(atomic.rmw_and,      V0, SPL_IR_ATOMIC_RMW_AND)        \
-    X(atomic.rmw_or,       V0, SPL_IR_ATOMIC_RMW_OR)         \
-    X(atomic.rmw_xor,      V0, SPL_IR_ATOMIC_RMW_XOR)        \
-    X(atomic.rmw_xchg,     V0, SPL_IR_ATOMIC_RMW_XCHG)       \
-    X(atomic.cmpxchg,      V0, SPL_IR_ATOMIC_CMPXCHG)        \
     X(control.select,      V0, SPL_IR_CONTROL_SELECT)        \
     X(control.br,          V0, SPL_IR_CONTROL_BR)            \
     X(control.jmp,         V0, SPL_IR_CONTROL_JMP)           \
@@ -74,7 +63,6 @@
     X(control.unreachable, V0, SPL_IR_CONTROL_UNREACHABLE)   \
     X(control.trap,        V0, SPL_IR_CONTROL_TRAP)          \
     X(dbg.breakpoint,      V0, SPL_IR_DBG_BREAKPOINT)        \
-    X(dbg.declare,         V0, SPL_IR_DBG_DECLARE)
 
 typedef enum {
 #ifdef X
@@ -95,6 +83,7 @@ typedef struct {
     spl_ir_kind_t kind;
     spl_dbg_node_t dbg;
     union {
+        spl_type_id_t tid;
         struct {
             spl_type_id_t tid;
             spl_ir_node_ref_t left;
@@ -149,9 +138,6 @@ typedef struct {
             spl_ir_node_ref_t val;
             spl_ir_node_ref_t size;
         } mem_set;
-        struct {
-            spl_ir_node_ref_t ordering;
-        } mem_fence;
         struct {
             spl_type_id_t tid;
             union {
@@ -236,7 +222,7 @@ typedef struct {
         SPL_IR_ATTR_NAKED,        /* 不实现 */
         SPL_IR_ATTR_NOINLINE,     /* 不实现 */
         SPL_IR_ATTR_ALWAYSINLINE, /* 不实现 */
-    };
+    } kind;
 } spl_ir_attr_t;
 typedef VEC(spl_ir_attr_t) spl_ir_attr_vec_t;
 
@@ -267,4 +253,145 @@ spl_ir_func_t *spl_ir_func(spl_ir_t *ir, spl_ir_func_ref_t fn_id);
 void spl_ir_dump(spl_ir_t *ir, const spl_type_t *ty);
 const char *spl_ir_kind_name(spl_ir_kind_t kind);
 
+typedef struct {
+    spl_ir_t ir;
+    spl_ir_func_ref_t current_fn;
+    spl_ir_node_ref_t current_node;
+    spl_ir_node_ref_vec_t reloc_label;
+    spl_dbg_node_t dbg;
+} spl_ir_builder_t;
+
+/* clang-format off */
+#define SPL_IR_BIN_ARITH_TABLE                              \
+    X(arith_add,     SPL_IR_ARITH_ADD)                      \
+    X(arith_sub,     SPL_IR_ARITH_SUB)                      \
+    X(arith_mul,     SPL_IR_ARITH_MUL)                      \
+    X(arith_div,     SPL_IR_ARITH_DIV)                      \
+    X(arith_rem,     SPL_IR_ARITH_REM)                      \
+    X(arith_and,     SPL_IR_ARITH_AND)                      \
+    X(arith_or,      SPL_IR_ARITH_OR)                       \
+    X(arith_xor,     SPL_IR_ARITH_XOR)                      \
+    X(arith_shl,     SPL_IR_ARITH_SHL)                      \
+    X(arith_shr,     SPL_IR_ARITH_SHR)
+#define SPL_IR_UN_ARITH_TABLE                               \
+    X(arith_neg,     SPL_IR_ARITH_NEG)                      \
+    X(arith_abs,     SPL_IR_ARITH_ABS)                      \
+    X(arith_not,     SPL_IR_ARITH_NOT)
+#define SPL_IR_CMP_TABLE                                    \
+    X(cmp_eq,        SPL_IR_CMP_EQ)                         \
+    X(cmp_ne,        SPL_IR_CMP_NE)                         \
+    X(cmp_lt,        SPL_IR_CMP_LT)                         \
+    X(cmp_le,        SPL_IR_CMP_LE)                         \
+    X(cmp_gt,        SPL_IR_CMP_GT)                         \
+    X(cmp_ge,        SPL_IR_CMP_GE)
+#define SPL_IR_CAST_TABLE                                   \
+    X(cast_trunc,     SPL_IR_CAST_TRUNC)                    \
+    X(cast_zext,      SPL_IR_CAST_ZEXT)                     \
+    X(cast_sext,      SPL_IR_CAST_SEXT)                     \
+    X(cast_fext,      SPL_IR_CAST_FEXT)                     \
+    X(cast_ftrunc,    SPL_IR_CAST_FTRUNC)                   \
+    X(cast_bitcast,   SPL_IR_CAST_BITCAST)                  \
+    X(cast_ptr2int,   SPL_IR_CAST_PTR2INT)                  \
+    X(cast_int2ptr,   SPL_IR_CAST_INT2PTR)                  \
+    X(cast_bool2int,  SPL_IR_CAST_BOOL2INT)                 \
+    X(case_int2float, SPL_IR_CASE_INT2FLOAT)                \
+    X(case_float2int, SPL_IR_CASE_FLOAT2INT)
+/* clang-format on */
+void spl_ir_builder_init(spl_ir_builder_t *b);
+void spl_ir_builder_drop(spl_ir_builder_t *b);
+
+spl_ir_func_ref_t spl_ir_builder_fn_new(spl_ir_builder_t *b, const char *name,
+                                        spl_type_id_t fn_tid);
+spl_ir_func_ref_t spl_ir_builder_cur_fn(const spl_ir_builder_t *b);
+void spl_ir_builder_set_dbg(spl_ir_builder_t *b, spl_dbg_node_t dbg);
+
+spl_ir_node_ref_t spl_ir_builder_block_new(spl_ir_builder_t *b);
+
+#define X(name, kind)                                                                              \
+    spl_ir_node_ref_t spl_ir_builder_##name(spl_ir_builder_t *b, spl_type_id_t tid,                \
+                                            spl_ir_node_ref_t left, spl_ir_node_ref_t right);
+SPL_IR_BIN_ARITH_TABLE
+#undef X
+
+#define X(name, kind)                                                                              \
+    spl_ir_node_ref_t spl_ir_builder_##name(spl_ir_builder_t *b, spl_type_id_t tid,                \
+                                            spl_ir_node_ref_t val);
+SPL_IR_UN_ARITH_TABLE
+#undef X
+
+#define X(name, kind)                                                                              \
+    spl_ir_node_ref_t spl_ir_builder_##name(spl_ir_builder_t *builder, spl_type_id_t tid,          \
+                                            spl_ir_node_ref_t a, spl_ir_node_ref_t b);
+SPL_IR_CMP_TABLE
+#undef X
+
+#define X(name, kind)                                                                              \
+    spl_ir_node_ref_t spl_ir_builder_##name(spl_ir_builder_t *b, spl_type_id_t from_tid,           \
+                                            spl_type_id_t to_tid, spl_ir_node_ref_t val);
+SPL_IR_CAST_TABLE
+#undef X
+
+spl_ir_node_ref_t spl_ir_builder_mem_alloca(spl_ir_builder_t *b, spl_type_id_t tid,
+                                            spl_ir_node_ref_t count);
+spl_ir_node_ref_t spl_ir_builder_mem_global_alloc(spl_ir_builder_t *b, spl_type_id_t tid,
+                                                  spl_ir_node_ref_t const_node);
+spl_ir_node_ref_t spl_ir_builder_mem_load(spl_ir_builder_t *b, spl_type_id_t tid,
+                                          spl_ir_node_ref_t ptr);
+spl_ir_node_ref_t spl_ir_builder_mem_store(spl_ir_builder_t *b, spl_type_id_t tid,
+                                           spl_ir_node_ref_t ptr, spl_ir_node_ref_t val);
+spl_ir_node_ref_t spl_ir_builder_mem_offset(spl_ir_builder_t *b, spl_type_id_t tid,
+                                            spl_ir_node_ref_t ptr, spl_ir_node_ref_t offset);
+spl_ir_node_ref_t spl_ir_builder_mem_field(spl_ir_builder_t *b, spl_type_id_t tid,
+                                           spl_ir_node_ref_t agg, usize field_idx);
+spl_ir_node_ref_t spl_ir_builder_mem_copy(spl_ir_builder_t *b, spl_type_id_t tid,
+                                          spl_ir_node_ref_t dst, spl_ir_node_ref_t src,
+                                          spl_ir_node_ref_t size);
+spl_ir_node_ref_t spl_ir_builder_mem_set(spl_ir_builder_t *b, spl_type_id_t tid,
+                                         spl_ir_node_ref_t dst, spl_ir_node_ref_t val,
+                                         spl_ir_node_ref_t size);
+
+spl_ir_node_ref_t spl_ir_builder_type_const_int(spl_ir_builder_t *b, spl_type_id_t tid, usize val);
+spl_ir_node_ref_t spl_ir_builder_type_const_float(spl_ir_builder_t *b, spl_type_id_t tid,
+                                                  double val);
+spl_ir_node_ref_t spl_ir_builder_type_const_cstr(spl_ir_builder_t *b, spl_type_id_t tid,
+                                                 const char *val);
+spl_ir_node_ref_t spl_ir_builder_type_const_char(spl_ir_builder_t *b, spl_type_id_t tid, char val);
+spl_ir_node_ref_t spl_ir_builder_type_const_fn(spl_ir_builder_t *b, spl_type_id_t tid,
+                                               spl_ir_func_ref_t val);
+spl_ir_node_ref_t spl_ir_builder_type_bitsizeof(spl_ir_builder_t *b, spl_type_id_t tid);
+spl_ir_node_ref_t spl_ir_builder_type_sizeof(spl_ir_builder_t *b, spl_type_id_t tid);
+spl_ir_node_ref_t spl_ir_builder_type_alignof(spl_ir_builder_t *b, spl_type_id_t tid);
+spl_ir_node_ref_t spl_ir_builder_type_offsetof(spl_ir_builder_t *b, spl_type_id_t tid,
+                                               spl_ir_node_ref_t field_idx);
+spl_ir_node_ref_t spl_ir_builder_type_field_count(spl_ir_builder_t *b, spl_type_id_t tid);
+
+spl_ir_node_ref_t spl_ir_builder_agg_construct(spl_ir_builder_t *b, spl_type_id_t tid);
+void spl_ir_builder_agg_construct_field(spl_ir_builder_t *b, spl_ir_node_ref_t node,
+                                        spl_ir_node_ref_t field);
+spl_ir_node_ref_t spl_ir_builder_agg_extract(spl_ir_builder_t *b, spl_type_id_t tid,
+                                             spl_type_id_t field_tid, usize field_idx,
+                                             spl_ir_node_ref_t val);
+spl_ir_node_ref_t spl_ir_builder_agg_insert(spl_ir_builder_t *b, spl_type_id_t tid, usize field_idx,
+                                            spl_ir_node_ref_t agg, spl_ir_node_ref_t field);
+
+spl_ir_node_ref_t spl_ir_builder_control_select(spl_ir_builder_t *b, spl_type_id_t tid,
+                                                spl_ir_node_ref_t cond, spl_ir_node_ref_t true_val,
+                                                spl_ir_node_ref_t false_val);
+spl_ir_node_ref_t spl_ir_builder_control_br(spl_ir_builder_t *b, spl_ir_node_ref_t cond,
+                                            spl_ir_node_ref_t true_label,
+                                            spl_ir_node_ref_t false_label);
+spl_ir_node_ref_t spl_ir_builder_control_jmp(spl_ir_builder_t *b, spl_ir_node_ref_t label);
+spl_ir_node_ref_t spl_ir_builder_control_call(spl_ir_builder_t *b, spl_type_id_t tid,
+                                              spl_ir_node_ref_t func);
+void spl_ir_builder_control_call_param(spl_ir_builder_t *b, spl_ir_node_ref_t node,
+                                       spl_ir_node_ref_t arg);
+spl_ir_node_ref_t spl_ir_builder_control_param(spl_ir_builder_t *b, spl_type_id_t tid,
+                                               spl_ir_node_ref_t idx);
+spl_ir_node_ref_t spl_ir_builder_control_ret(spl_ir_builder_t *b, spl_type_id_t tid,
+                                             spl_ir_node_ref_t val);
+spl_ir_node_ref_t spl_ir_builder_control_unreachable(spl_ir_builder_t *b);
+spl_ir_node_ref_t spl_ir_builder_control_trap(spl_ir_builder_t *b);
+
+spl_ir_node_ref_t spl_ir_builder_dbg_breakpoint(spl_ir_builder_t *b);
+
 #endif /* __SPL_IR_H__ */

stage1/spl_sema.h

@@ -4,32 +4,53 @@
 #include "spl_ast.h"
 #include "spl_type.h"
 
+typedef enum {
+    SPL_SYMBOL_KIND_ERROR,
+    SPL_SYMBOL_KIND_VAR,
+    SPL_SYMBOL_KIND_MEMBER,
+    SPL_SYMBOL_KIND_FN,
+    SPL_SYMBOL_KIND_TYPE,
+} spl_symbol_kind_t;
+typedef struct {
+    const char *name;
+    spl_symbol_kind_t kind;
+    spl_def_id_t node;
+} spl_symbol_t;
+
 typedef usize spl_scope_id_t; /* 0 is error */
 typedef struct {
     spl_scope_id_t parent;
-    MAP(const char *, spl_def_id_t) symbols;
+    MAP(const char *, spl_symbol_t) symbols;
 } spl_scope_node_t;
 typedef VEC(spl_scope_node_t) spl_scope_node_vec_t;
-
 typedef struct {
-    spl_ast_t *ast;
-    spl_type_t type;
     spl_scope_node_vec_t scopes;
     spl_scope_id_t root_scope;
     spl_scope_id_t current_scope;
-    spl_def_id_t root_def;
+} spl_scope_t;
+
+void spl_scope_init(spl_scope_t *scope);
+void spl_scope_drop(spl_scope_t *scope);
+
+spl_scope_id_t spl_scope_alloc(spl_scope_t *scope);
+bool spl_scope_insert(spl_scope_t *scope, spl_scope_id_t id, spl_symbol_t symbol);
+typedef VEC(const char *) spl_symbol_path_t;
+bool spl_scope_find(spl_scope_t *scop, const char *symbol_name, spl_symbol_t *out);
+
+typedef struct {
+    spl_ast_t *ast;
+    spl_type_t *type;
+    spl_scope_t *scope;
+    spl_symbol_t root;
     int error_count;
+    spl_scope_id_t store_scope;
+    spl_type_id_t current_fn_ret_tid; // TODO
 } spl_sema_t;
 
-void spl_sema_init(spl_sema_t *sema);
+void spl_sema_init(spl_sema_t *sema, spl_ast_t *ast, spl_type_t *type, spl_scope_t *scope);
 void spl_sema_drop(spl_sema_t *sema);
-void spl_sema_run(spl_sema_t *sema);
-void spl_sema_check(spl_sema_t *sema);
 
-spl_scope_id_t spl_sema_scope_alloc(spl_sema_t *sema);
-bool spl_sema_scope_insert(spl_sema_t *sema, spl_scope_id_t id, const char *symbol_name,
-                           spl_def_id_t symbol_val);
-typedef VEC(const char *) spl_symbol_path_t;
-spl_def_id_t spl_sema_scope_find(spl_sema_t *sema, spl_symbol_path_t path);
+void spl_sema_run(spl_sema_t *sema);
+void spl_sema_dump(spl_sema_t *sema);
 
 #endif /* __SPL_SEMA_H__ */

stage1/spl_type.c

@@ -1,75 +1,356 @@
 #include "spl_type.h"
 
-static usize spl_type_hash(spl_type_node_t n) { return 0; }
-static int spl_type_eq(spl_type_node_t n1, spl_type_node_t n2) { return 1; }
-
-spl_type_id_t spl_type_node_push(spl_type_t *type, spl_type_node_t type_node) {
-    spl_type_id_t ret = 0;
-    int ok = map_get(type->type_map, type_node, &ret);
-    if (ok) {
-        return ret;
+static usize spl_type_hash(spl_type_node_t n) {
+    usize hash = n.kind << 20;
+    switch (n.kind) {
+    case SPL_TYPE_ERROR:
+    case SPL_TYPE_VOID:
+    case SPL_TYPE_BOOL:
+    case SPL_TYPE_UNDEFINED:
+    case SPL_TYPE_NULL:
+        break;
+    case SPL_TYPE_INT:
+        hash += n.int_type.bits + n.int_type.is_signed;
+        break;
+    case SPL_TYPE_FLOAT:
+        hash += n.float_type.bits;
+        break;
+    case SPL_TYPE_PTR:
+        hash += n.ptr_pointee;
+        break;
+    case SPL_TYPE_SLICE:
+        hash += n.slice_element;
+        break;
+    case SPL_TYPE_RANGE:
+        hash += n.range_element;
+        break;
+    case SPL_TYPE_ARRAY:
+        hash += n.array_type.element + n.array_type.len;
+        break;
+    case SPL_TYPE_STRUCT:
+    case SPL_TYPE_UNION:
+        hash += n.layout.mode * 31 + n.layout.fixed_align_bits;
+        vec_for(n.agg_members, i) {
+            hash += (n.agg_members.data[i].name ? map_hash_str(n.agg_members.data[i].name) : 0) +
+                    n.agg_members.data[i].type_id;
+        }
+        break;
+    case SPL_TYPE_ENUM:
+        hash += n.layout.mode * 31 + n.layout.fixed_align_bits;
+        hash += n.adt_type.tag_type;
+        vec_for(n.adt_type.variants, i) {
+            hash +=
+                (n.adt_type.variants.data[i].name ? map_hash_str(n.adt_type.variants.data[i].name)
+                                                  : 0) +
+                n.adt_type.variants.data[i].type_id;
+        }
+        break;
+    case SPL_TYPE_FN:
+        hash += n.fn_type.ret;
+        vec_for(n.fn_type.params, i) { hash += vec_at(n.fn_type.params, i); }
+        break;
+    default: // TODO
+        break;
     }
-    map_put(type->type_map, type_node, vec_size(type->type_table));
-    vec_push(type->type_table, type_node);
-    return vec_size(type->type_table) - 1;
+    return hash;
 }
 
-spl_type_id_t spl_type_def_alloc(spl_type_t *type) {
-    vec_push(type->def_table, (spl_def_node_t){0});
+static int spl_type_eq(spl_type_node_t n1, spl_type_node_t n2) {
+    if (n1.kind != n2.kind)
+        return 1;
+    switch (n1.kind) {
+    case SPL_TYPE_ERROR:
+    case SPL_TYPE_VOID:
+    case SPL_TYPE_BOOL:
+    case SPL_TYPE_UNDEFINED:
+    case SPL_TYPE_NULL:
+        break;
+    case SPL_TYPE_INT:
+        if (n1.int_type.bits != n2.int_type.bits)
+            return 1;
+        if (n1.int_type.is_signed != n2.int_type.is_signed)
+            return 1;
+        break;
+    case SPL_TYPE_FLOAT:
+        if (n1.float_type.bits != n2.float_type.bits)
+            return 1;
+        break;
+    case SPL_TYPE_PTR:
+        if (n1.ptr_pointee != n2.ptr_pointee)
+            return 1;
+        break;
+    case SPL_TYPE_SLICE:
+        if (n1.slice_element != n2.slice_element)
+            return 1;
+        break;
+    case SPL_TYPE_RANGE:
+        if (n1.range_element != n2.range_element)
+            return 1;
+        break;
+    case SPL_TYPE_ARRAY:
+        if (n1.array_type.element != n2.array_type.element)
+            return 1;
+        if (n1.array_type.len != n2.array_type.len)
+            return 1;
+        break;
+    case SPL_TYPE_STRUCT:
+    case SPL_TYPE_UNION:
+        if (n1.layout.mode != n2.layout.mode ||
+            n1.layout.fixed_align_bits != n2.layout.fixed_align_bits)
+            return 1;
+        if (vec_size(n1.agg_members) != vec_size(n2.agg_members))
+            return 1;
+        vec_for(n1.agg_members, i) {
+            const char *na = n1.agg_members.data[i].name;
+            const char *nb = n2.agg_members.data[i].name;
+            if (strcmp(na ? na : "", nb ? nb : "") != 0)
+                return 1;
+            if (n1.agg_members.data[i].type_id != n2.agg_members.data[i].type_id)
+                return 1;
+        }
+        break;
+    case SPL_TYPE_ENUM:
+        if (n1.layout.mode != n2.layout.mode ||
+            n1.layout.fixed_align_bits != n2.layout.fixed_align_bits)
+            return 1;
+        if (n1.adt_type.tag_type != n2.adt_type.tag_type)
+            return 1;
+        if (vec_size(n1.adt_type.variants) != vec_size(n2.adt_type.variants))
+            return 1;
+        vec_for(n1.adt_type.variants, i) {
+            const char *na = n1.adt_type.variants.data[i].name;
+            const char *nb = n2.adt_type.variants.data[i].name;
+            if (strcmp(na ? na : "", nb ? nb : "") != 0)
+                return 1;
+            if (n1.adt_type.variants.data[i].type_id != n2.adt_type.variants.data[i].type_id)
+                return 1;
+        }
+        break;
+    case SPL_TYPE_FN:
+        if (n1.fn_type.ret != n2.fn_type.ret)
+            return 1;
+        if (vec_size(n1.fn_type.params) != vec_size(n2.fn_type.params))
+            return 1;
+        vec_for(n1.fn_type.params, i) {
+            if (vec_at(n1.fn_type.params, i) != vec_at(n2.fn_type.params, i))
+                return 1;
+        }
+        break;
+    default: // TODO
+        return 1;
+    }
+    return 0;
+}
+
+static usize spl_type_map_hash(usize k) { return MAP_HASH_INT(k); }
+
+static int spl_type_map_cmp(usize a, usize b) { return MAP_CMP_INT(a, b); }
+
+static void spl_type_node_free_vecs(spl_type_node_t *tn) {
+    switch (tn->kind) {
+    case SPL_TYPE_STRUCT:
+    case SPL_TYPE_UNION:
+        vec_free(tn->agg_members);
+        break;
+    case SPL_TYPE_ENUM:
+        vec_free(tn->adt_type.variants);
+        break;
+    case SPL_TYPE_FN:
+        vec_free(tn->fn_type.params);
+        break;
+    default:
+        break;
+    }
+}
+
+spl_type_id_t spl_type_node_push(spl_type_t *type, spl_type_node_t *tn) {
+    usize hash = spl_type_hash(*tn);
+    usize idx = (usize)-1;
+    spl_type_id_t cand = 0;
+    while (map_get_continue(type->type_map, hash, &cand, &idx)) {
+        spl_type_node_t *existing = &vec_at(type->type_table, cand);
+        if (spl_type_eq(*existing, *tn) == 0) {
+            spl_type_node_free_vecs(tn);
+            return cand;
+        }
+    }
+    spl_type_id_t id = vec_size(type->type_table);
+    vec_push(type->type_table, *tn);
+    map_put_new(type->type_map, hash, id);
+    tn->kind = SPL_TYPE_ERROR;
+    return id;
+}
+
+static const spl_type_member_t *spl_type_agg_member_find(spl_type_t *type, spl_type_id_t tid,
+                                                         const char *name, usize *out_idx) {
+    spl_type_node_t *n = spl_type_node(type, tid);
+    if (!n || !name) {
+        return NULL;
+    }
+    if (n->kind == SPL_TYPE_STRUCT || n->kind == SPL_TYPE_UNION) {
+        vec_for(n->agg_members, i) {
+            const char *mn = n->agg_members.data[i].name;
+            if (strcmp(mn ? mn : "", name) == 0) {
+                if (out_idx) {
+                    *out_idx = i;
+                }
+                return &n->agg_members.data[i];
+            }
+        }
+    } else if (n->kind == SPL_TYPE_ENUM) {
+        vec_for(n->adt_type.variants, i) {
+            const char *mn = n->adt_type.variants.data[i].name;
+            if (strcmp(mn ? mn : "", name) == 0) {
+                if (out_idx) {
+                    *out_idx = i;
+                }
+                return &n->adt_type.variants.data[i];
+            }
+        }
+    }
+    return NULL;
+}
+
+const spl_type_member_t *spl_type_agg_member(spl_type_t *type, spl_type_id_t tid,
+                                             const char *name) {
+    return spl_type_agg_member_find(type, tid, name, NULL);
+}
+
+bool spl_type_agg_member_idx(spl_type_t *type, spl_type_id_t tid, const char *name,
+                             usize *out_idx) {
+    return spl_type_agg_member_find(type, tid, name, out_idx) != NULL;
+}
+
+spl_type_id_t spl_type_builder_simple(spl_type_t *type, spl_type_node_kind_t kind) {
+    spl_type_node_t tn = {0};
+    tn.kind = kind;
+    return spl_type_node_push(type, &tn);
+}
+
+spl_type_id_t spl_type_builder_int(spl_type_t *type, usize bits, int is_signed) {
+    spl_type_node_t tn = {0};
+    tn.kind = SPL_TYPE_INT;
+    tn.int_type.bits = bits;
+    tn.int_type.is_signed = is_signed;
+    return spl_type_node_push(type, &tn);
+}
+
+spl_type_id_t spl_type_builder_float(spl_type_t *type, usize bits) {
+    spl_type_node_t tn = {0};
+    tn.kind = SPL_TYPE_FLOAT;
+    tn.float_type.bits = bits;
+    return spl_type_node_push(type, &tn);
+}
+
+spl_type_id_t spl_type_builder_ptr(spl_type_t *type, spl_type_id_t pointee) {
+    spl_type_node_t tn = {0};
+    tn.kind = SPL_TYPE_PTR;
+    tn.ptr_pointee = pointee;
+    return spl_type_node_push(type, &tn);
+}
+
+spl_type_id_t spl_type_builder_slice(spl_type_t *type, spl_type_id_t element) {
+    spl_type_node_t tn = {0};
+    tn.kind = SPL_TYPE_SLICE;
+    tn.slice_element = element;
+    return spl_type_node_push(type, &tn);
+}
+
+spl_type_id_t spl_type_builder_range(spl_type_t *type, spl_type_id_t element) {
+    spl_type_node_t tn = {0};
+    tn.kind = SPL_TYPE_RANGE;
+    tn.range_element = element;
+    return spl_type_node_push(type, &tn);
+}
+
+spl_type_id_t spl_type_builder_array(spl_type_t *type, spl_type_id_t element, usize len) {
+    spl_type_node_t tn = {0};
+    tn.kind = SPL_TYPE_ARRAY;
+    tn.array_type.element = element;
+    tn.array_type.len = len;
+    return spl_type_node_push(type, &tn);
+}
+
+spl_type_id_t spl_type_builder_fn(spl_type_t *type, spl_type_id_t ret, const spl_type_id_t *params,
+                                  usize nparams) {
+    spl_type_node_t tn = {0};
+    tn.kind = SPL_TYPE_FN;
+    vec_init(tn.fn_type.params);
+    for (usize i = 0; i < nparams; i++) {
+        vec_push(tn.fn_type.params, params[i]);
+    }
+    tn.fn_type.ret = ret;
+    return spl_type_node_push(type, &tn);
+}
+
+spl_type_id_t spl_type_builder_agg(spl_type_t *type, spl_type_node_kind_t tk,
+                                   const spl_type_member_t *members, usize nmembers,
+                                   spl_type_id_t tag_type, spl_type_layout_t layout) {
+    spl_type_node_t tn = {0};
+    tn.kind = tk;
+    tn.layout = layout;
+    if (tk == SPL_TYPE_ENUM) {
+        vec_init(tn.adt_type.variants);
+        for (usize i = 0; i < nmembers; i++) {
+            vec_push(tn.adt_type.variants, members[i]);
+        }
+        tn.adt_type.tag_type = tag_type;
+    } else {
+        vec_init(tn.agg_members);
+        for (usize i = 0; i < nmembers; i++) {
+            vec_push(tn.agg_members, members[i]);
+        }
+    }
+    return spl_type_node_push(type, &tn);
+}
+
+spl_def_id_t spl_type_def_alloc(spl_type_t *type, spl_def_node_kind_t kind, const char *name,
+                                spl_dbg_node_t dbg, usize ast_node_ref) {
+    spl_def_node_t n = {0};
+    n.kind = kind;
+    n.name = name;
+    n.dbg_node = dbg;
+    n.ast_node_ref = ast_node_ref;
+    vec_push(type->def_table, n);
     return vec_size(type->def_table) - 1;
 }
 
+void spl_type_def_resolve(spl_type_t *type, spl_def_id_t id, spl_type_id_t tid) {
+    spl_def_node_t *d = spl_type_def(type, id);
+    if (!d) {
+        return;
+    }
+    d->type_id = tid;
+}
+
+spl_type_id_t spl_type_fn_param_tid(spl_type_t *type, spl_type_id_t fn_tid, usize idx) {
+    spl_type_node_t *n = spl_type_node(type, fn_tid);
+    if (!n || n->kind != SPL_TYPE_FN || idx >= vec_size(n->fn_type.params))
+        return 0;
+    return vec_at(n->fn_type.params, idx);
+}
+
 void spl_type_init(spl_type_t *type) {
     vec_init(type->type_table);
     vec_push(type->type_table, (spl_type_node_t){0});
-    map_init(type->type_map, spl_type_hash, spl_type_eq);
+    map_init(type->type_map, spl_type_map_hash, spl_type_map_cmp);
     vec_init(type->def_table);
     vec_push(type->def_table, (spl_def_node_t){0});
 
-    spl_type_node_push(type, (spl_type_node_t){.kind = SPL_TYPE_VOID});
-    spl_type_node_push(type, (spl_type_node_t){.kind = SPL_TYPE_BOOL});
-    spl_type_node_push(type, (spl_type_node_t){
-                                 .kind = SPL_TYPE_INT,
-                                 .int_type.bits = 8,
-                                 .int_type.is_signed = false,
-                             });
-    spl_type_node_push(type, (spl_type_node_t){
-                                 .kind = SPL_TYPE_INT,
-                                 .int_type.bits = 16,
-                                 .int_type.is_signed = false,
-                             });
-    spl_type_node_push(type, (spl_type_node_t){
-                                 .kind = SPL_TYPE_INT,
-                                 .int_type.bits = 32,
-                                 .int_type.is_signed = false,
-                             });
-    spl_type_node_push(type, (spl_type_node_t){
-                                 .kind = SPL_TYPE_INT,
-                                 .int_type.bits = 64,
-                                 .int_type.is_signed = false,
-                             });
-    spl_type_node_push(type, (spl_type_node_t){
-                                 .kind = SPL_TYPE_INT,
-                                 .int_type.bits = 8,
-                                 .int_type.is_signed = true,
-                             });
-    spl_type_node_push(type, (spl_type_node_t){
-                                 .kind = SPL_TYPE_INT,
-                                 .int_type.bits = 16,
-                                 .int_type.is_signed = true,
-                             });
-    spl_type_node_push(type, (spl_type_node_t){
-                                 .kind = SPL_TYPE_INT,
-                                 .int_type.bits = 32,
-                                 .int_type.is_signed = true,
-                             });
-    spl_type_node_push(type, (spl_type_node_t){
-                                 .kind = SPL_TYPE_INT,
-                                 .int_type.bits = 64,
-                                 .int_type.is_signed = true,
-                             });
-    spl_type_node_push(type, (spl_type_node_t){.kind = SPL_TYPE_FLOAT, .float_type.bits = 32});
-    spl_type_node_push(type, (spl_type_node_t){.kind = SPL_TYPE_FLOAT, .float_type.bits = 64});
+    spl_type_builder_simple(type, SPL_TYPE_VOID);
+    spl_type_builder_simple(type, SPL_TYPE_BOOL);
+    spl_type_builder_simple(type, SPL_TYPE_UNDEFINED);
+    spl_type_builder_simple(type, SPL_TYPE_NULL);
+    spl_type_builder_int(type, 8, false);
+    spl_type_builder_int(type, 16, false);
+    spl_type_builder_int(type, 32, false);
+    spl_type_builder_int(type, 64, false);
+    spl_type_builder_int(type, 8, true);
+    spl_type_builder_int(type, 16, true);
+    spl_type_builder_int(type, 32, true);
+    spl_type_builder_int(type, 64, true);
+    spl_type_builder_float(type, 32);
+    spl_type_builder_float(type, 64);
 }
 
 void spl_type_drop(spl_type_t *type) {
@@ -78,7 +359,7 @@ void spl_type_drop(spl_type_t *type) {
         switch (n->kind) {
         case SPL_TYPE_STRUCT:
         case SPL_TYPE_UNION:
-            vec_free(n->agg_field_types);
+            vec_free(n->agg_members);
             break;
         case SPL_TYPE_ENUM:
             vec_free(n->adt_type.variants);
@@ -91,20 +372,6 @@ void spl_type_drop(spl_type_t *type) {
         }
     }
 
-    vec_for(type->def_table, i) {
-        spl_def_node_t *d = &vec_at(type->def_table, i);
-        switch (d->kind) {
-        case SPL_DEF_AGG:
-            vec_free(d->agg_def);
-            break;
-        case SPL_DEF_FN_PARAMS:
-            vec_free(d->fn_params_def);
-            break;
-        default:
-            break;
-        }
-    }
-
     map_free(type->type_map);
     vec_free(type->type_table);
     vec_free(type->def_table);
@@ -125,7 +392,7 @@ spl_def_node_t *spl_type_def(spl_type_t *type, spl_def_id_t id) {
 void spl_type_pure_dump(spl_type_t *type, spl_type_id_t id) {
     spl_type_node_t *n = spl_type_node(type, id);
     if (!n) {
-        printf("(err)");
+        printf("none");
         return;
     }
     switch (n->kind) {
@@ -138,11 +405,25 @@ void spl_type_pure_dump(spl_type_t *type, spl_type_id_t id) {
     case SPL_TYPE_BOOL:
         printf("bool");
         break;
+    case SPL_TYPE_UNDEFINED:
+        printf("undefined");
+        break;
+    case SPL_TYPE_NULL:
+        printf("null");
+        break;
     case SPL_TYPE_INT:
-        printf("%s%zu", n->int_type.is_signed ? "i" : "u", n->int_type.bits);
+        if (n->int_type.bits == 0) {
+            printf("comptime_int");
+        } else {
+            printf("%s%zu", n->int_type.is_signed ? "i" : "u", n->int_type.bits);
+        }
         break;
     case SPL_TYPE_FLOAT:
-        printf("f%zu", n->float_type.bits);
+        if (n->float_type.bits == 0) {
+            printf("comptime_float");
+        } else {
+            printf("f%zu", n->float_type.bits);
+        }
         break;
     case SPL_TYPE_PTR:
         printf("*");
@@ -162,20 +443,48 @@ void spl_type_pure_dump(spl_type_t *type, spl_type_id_t id) {
         spl_type_pure_dump(type, n->array_type.element);
         break;
     case SPL_TYPE_STRUCT:
-        printf("struct{%zu fields}", n->agg_field_types.size);
+        printf("struct{");
+        vec_for(n->agg_members, i) {
+            if (i) {
+                printf(", ");
+            }
+            spl_type_member_t *member = &vec_at(n->agg_members, i);
+            printf("%s:", member->name ? member->name : "?");
+            spl_type_pure_dump(type, member->type_id);
+        }
+        printf("}");
         break;
     case SPL_TYPE_UNION:
-        printf("union{%zu fields}", n->agg_field_types.size);
+        printf("union{");
+        vec_for(n->agg_members, i) {
+            if (i) {
+                printf(", ");
+            }
+            spl_type_member_t *member = &vec_at(n->agg_members, i);
+            printf("%s:", member->name ? member->name : "?");
+            spl_type_pure_dump(type, member->type_id);
+        }
+        printf("}");
         break;
     case SPL_TYPE_ENUM:
-        printf("enum{%zu variants}", n->adt_type.variants.size);
+        printf("enum{"); // TODO tag_type
+        vec_for(n->adt_type.variants, i) {
+            if (i) {
+                printf(", ");
+            }
+            spl_type_member_t *member = &vec_at(n->adt_type.variants, i);
+            printf("%s:", member->name ? member->name : "?");
+            spl_type_pure_dump(type, member->type_id);
+        }
+        printf("}");
         break;
     case SPL_TYPE_FN: {
         printf("fn(");
-        for (usize i = 0; i < n->fn_type.params.size; i++) {
-            if (i)
+        vec_for(n->fn_type.params, i) {
+            if (i) {
                 printf(",");
-            spl_type_pure_dump(type, n->fn_type.params.data[i]);
+            }
+            spl_type_pure_dump(type, vec_at(n->fn_type.params, i));
         }
         printf(") -> ");
         spl_type_pure_dump(type, n->fn_type.ret);
@@ -190,7 +499,7 @@ void spl_type_pure_dump(spl_type_t *type, spl_type_id_t id) {
 void spl_type_def_dump(spl_type_t *type, spl_def_id_t id) {
     spl_def_node_t *node = spl_type_def(type, id);
     if (!node) {
-        printf("(err)");
+        printf("none");
         return;
     }
 
@@ -209,7 +518,7 @@ void spl_type_def_dump(spl_type_t *type, spl_def_id_t id) {
         def_kind_name = "var";
         break;
     case SPL_DEF_FN_PARAMS:
-        def_kind_name = "params";
+        def_kind_name = "fn_params";
         break;
     case SPL_DEF_AGG:
         def_kind_name = "agg";
@@ -221,37 +530,6 @@ void spl_type_def_dump(spl_type_t *type, spl_def_id_t id) {
         def_kind_name = "sametypes";
         break;
     }
-    printf("kind=%s type_id=%zu", def_kind_name, node->type_id);
-    switch (node->kind) {
-    case SPL_DEF_VAR:
-        printf(" var=%s", node->var_def.name ? node->var_def.name : "?");
-        break;
-    case SPL_DEF_MEMBER:
-        printf(" member=%s", node->var_def.name ? node->var_def.name : "?");
-        break;
-    case SPL_DEF_ALIAS:
-    case SPL_DEF_DISTINCT:
-        printf(" type=%s", node->type_def.name ? node->type_def.name : "?");
-        break;
-    case SPL_DEF_AGG:
-        printf(" agg{");
-        for (usize i = 0; i < node->agg_def.size; i++) {
-            if (i)
-                printf(",");
-            printf("%s", node->agg_def.data[i].name ? node->agg_def.data[i].name : "?");
-        }
-        printf("}");
-        break;
-    case SPL_DEF_FN_PARAMS:
-        printf(" fn(");
-        for (usize i = 0; i < node->fn_params_def.size; i++) {
-            if (i)
-                printf(",");
-            printf("%s", node->fn_params_def.data[i].name ? node->fn_params_def.data[i].name : "?");
-        }
-        printf(")");
-        break;
-    default:
-        break;
-    }
+    printf("kind=%s type_id=%zu name=%s", def_kind_name, node->type_id,
+           node->name ? node->name : "?");
 }

stage1/spl_type.h

@@ -2,33 +2,53 @@
 #define __SPL_TYPE_H__
 
 #include "../stage0/include/utils.h"
+#include "spl_dbg.h"
 
 typedef usize spl_type_id_t; /* 0 is error */
 typedef VEC(spl_type_id_t) spl_type_id_vec_t;
 typedef usize spl_def_id_t; /* 0 is error */
 typedef VEC(spl_def_id_t) spl_def_id_vec_t;
 
+typedef enum {
+    SPL_TYPE_ERROR,
+    SPL_TYPE_VOID,
+    SPL_TYPE_BOOL,
+    SPL_TYPE_INT,
+    SPL_TYPE_FLOAT,
+    SPL_TYPE_UNDEFINED, // 自动配对任意类型的字面量，值为 undefined
+    SPL_TYPE_NULL,      // 自动配对指针/切片的字面量，值为 0
+    SPL_TYPE_PTR,
+    SPL_TYPE_SLICE, // 未来拥有泛型后删除
+    SPL_TYPE_RANGE, // 未来拥有泛型后删除
+    SPL_TYPE_ARRAY,
+    SPL_TYPE_STRUCT,
+    SPL_TYPE_UNION,
+    SPL_TYPE_ENUM,
+    SPL_TYPE_FN,
+    SPL_TYPE_ID,
+} spl_type_node_kind_t;
+
+typedef enum {
+    SPL_TYPE_LAYOUT_AUTO,
+    SPL_TYPE_LAYOUT_EXTERN_C,
+    SPL_TYPE_LAYOUT_PACKED,
+} spl_type_layout_mode_t;
+
 typedef struct {
-    enum {
-        SPL_TYPE_ERROR,
-        SPL_TYPE_VOID,
-        SPL_TYPE_BOOL,
-        SPL_TYPE_INT,
-        SPL_TYPE_FLOAT,
-        SPL_TYPE_PTR,
-        SPL_TYPE_SLICE, // 未来拥有泛型后删除
-        SPL_TYPE_RANGE, // 未来拥有泛型后删除
-        SPL_TYPE_ARRAY,
-        SPL_TYPE_STRUCT,
-        SPL_TYPE_UNION,
-        SPL_TYPE_ENUM,
-        SPL_TYPE_FN,
-        SPL_TYPE_ID,
-    } kind;
-    struct {
-        enum { AUTO, EXTERN_C, PACKED } mode;
-        usize fixed_align_bits;
-    } layout;
+    spl_type_layout_mode_t mode;
+    usize fixed_align_bits;
+} spl_type_layout_t;
+
+typedef struct {
+    const char *name;
+    spl_type_id_t type_id;
+    usize ast_node_ref; // 回 AST member_decl 节点引用 dbg/IDE 定位用 不参与 hash
+} spl_type_member_t;
+typedef VEC(spl_type_member_t) spl_type_member_vec_t;
+
+typedef struct {
+    spl_type_node_kind_t kind;
+    spl_type_layout_t layout;
     union {
         struct {
             usize bits;
@@ -44,9 +64,9 @@ typedef struct {
             spl_type_id_t element;
             usize len;
         } array_type;
-        spl_type_id_vec_t agg_field_types;
+        spl_type_member_vec_t agg_members;
         struct {
-            spl_type_id_vec_t variants;
+            spl_type_member_vec_t variants;
             spl_type_id_t tag_type;
         } adt_type; // ADT
         struct {
@@ -58,39 +78,29 @@ typedef struct {
 } spl_type_node_t;
 typedef VEC(spl_type_node_t) spl_type_node_vec_t;
 
-typedef struct {
-    const char *name;
-    spl_def_id_t def_id;
-    spl_type_id_t type_id;
-    usize scope_id;
-} spl_var_def_t;
-typedef VEC(spl_var_def_t) spl_var_def_vec_t;
+typedef enum {
+    SPL_DEF_ERROR,
+    SPL_DEF_SCALAR,
+    SPL_DEF_MEMBER,
+    SPL_DEF_VAR,
+    SPL_DEF_FN_PARAMS,
+    SPL_DEF_AGG,      // include enum variants
+    SPL_DEF_DISTINCT, // newtype
+    SPL_DEF_ALIAS,    // sametypes
+} spl_def_node_kind_t;
 
 typedef struct {
-    enum {
-        SPL_DEF_ERROR,
-        SPL_DEF_SCALAR,
-        SPL_DEF_MEMBER,
-        SPL_DEF_VAR,
-        SPL_DEF_FN_PARAMS,
-        SPL_DEF_AGG,      // include enum variants
-        SPL_DEF_DISTINCT, // newtype
-        SPL_DEF_ALIAS,    // sametypes
-    } kind;
+    spl_def_node_kind_t kind;
+    const char *name;
+    spl_dbg_node_t dbg_node;
     spl_type_id_t type_id;
-    union {
-        spl_var_def_t var_def;
-        spl_var_def_vec_t agg_def; // include enum variants
-        spl_var_def_vec_t fn_params_def;
-        spl_var_def_t type_def;
-    };
+    usize ast_node_ref;
 } spl_def_node_t;
 typedef VEC(spl_def_node_t) spl_def_node_vec_t;
 
-typedef MAP(spl_type_node_t, usize) spl_type_node_map_t;
+typedef MAP(usize, spl_type_id_t) spl_type_node_map_t;
 typedef struct {
     spl_type_node_vec_t type_table;
-    // TODO hashconsing
     spl_type_node_map_t type_map;
     spl_def_node_vec_t def_table;
 } spl_type_t;
@@ -104,7 +114,25 @@ void spl_type_pure_dump(spl_type_t *type, spl_type_id_t id);
 spl_type_node_t *spl_type_node(spl_type_t *type, spl_type_id_t id);
 spl_def_node_t *spl_type_def(spl_type_t *type, spl_def_id_t id);
 
-spl_type_id_t spl_type_def_alloc(spl_type_t *type);
-spl_type_id_t spl_type_node_push(spl_type_t *type, spl_type_node_t type_node);
+spl_type_id_t spl_type_def_alloc(spl_type_t *type, spl_def_node_kind_t kind, const char *name,
+                                 spl_dbg_node_t dbg, usize ast_node_ref);
+void spl_type_def_resolve(spl_type_t *type, spl_def_id_t id, spl_type_id_t tid);
 
+spl_type_id_t spl_type_fn_param_tid(spl_type_t *type, spl_type_id_t fn_tid, usize idx);
+const spl_type_member_t *spl_type_agg_member(spl_type_t *type, spl_type_id_t tid, const char *name);
+bool spl_type_agg_member_idx(spl_type_t *type, spl_type_id_t tid, const char *name, usize *out_idx);
+
+spl_type_id_t spl_type_node_push(spl_type_t *type, spl_type_node_t *type_node);
+spl_type_id_t spl_type_builder_simple(spl_type_t *type, spl_type_node_kind_t kind);
+spl_type_id_t spl_type_builder_int(spl_type_t *type, usize bits, int is_signed);
+spl_type_id_t spl_type_builder_float(spl_type_t *type, usize bits);
+spl_type_id_t spl_type_builder_ptr(spl_type_t *type, spl_type_id_t pointee);
+spl_type_id_t spl_type_builder_slice(spl_type_t *type, spl_type_id_t element);
+spl_type_id_t spl_type_builder_range(spl_type_t *type, spl_type_id_t element);
+spl_type_id_t spl_type_builder_array(spl_type_t *type, spl_type_id_t element, usize len);
+spl_type_id_t spl_type_builder_fn(spl_type_t *type, spl_type_id_t ret, const spl_type_id_t *params,
+                                  usize nparams);
+spl_type_id_t spl_type_builder_agg(spl_type_t *type, spl_type_node_kind_t tk,
+                                   const spl_type_member_t *members, usize nmembers,
+                                   spl_type_id_t tag_type, spl_type_layout_t layout);
 #endif /* __SPL_TYPE_H__ */

stage1/splc0.c

@@ -12,14 +12,23 @@
 
 #include "spl_ast.h"
 #include "spl_ast2ir.h"
-#include "spl_ir2vm.h"
+// #include "spl_ir2vm.h"
 #include "spl_lexer.h"
 #include "spl_sema.h"
 #include "spl_tok.h"
 
+typedef enum {
+    DUMP_NONE,
+    DUMP_TOKEN,
+    DUMP_AST,
+    DUMP_SEMA,
+    DUMP_IR,
+} dump_t;
+
 static char *read_file(const char *path, long *out_len) {
     FILE *f = fopen(path, "rb");
     if (!f) {
+        fprintf(stderr, "path `%s` ", path);
         perror("fopen");
         return NULL;
     }
@@ -47,8 +56,7 @@ static const char *const tok_type_names[] = {
 #undef X
 };
 
-static void dump_tokens(const char *src, const char *fname) {
-    spl_tok_vec_t toks = spl_lex(src, fname);
+static void dump_tokens(spl_tok_vec_t toks) {
     printf("tokens got (%zu)\n", toks.size);
     for (usize i = 0; i < toks.size; i++) {
         const spl_tok_t *t = &toks.data[i];
@@ -58,180 +66,66 @@ static void dump_tokens(const char *src, const char *fname) {
     vec_free(toks);
 }
 
-static void dump_ast(const char *src, const char *fname) {
+static int compile_spl(const char *src, const char *fname, const char *outpath, int gen_debug,
+                       dump_t dump) {
     spl_tok_vec_t toks = spl_lex(src, fname);
+    if (dump == DUMP_TOKEN) {
+        dump_tokens(toks);
+        // memory leak
+        return 0;
+    }
     spl_ast_t ast;
     spl_ast_init(&ast, &toks);
     spl_ast_prase(&ast);
     if (ast.parsed < 0) {
-        printf("parse failed, skip AST dump\n");
-        spl_ast_drop(&ast);
-        return;
-    }
-    spl_ast_valid(&ast);
-    spl_ast_dump(&ast, ast.root);
-    spl_ast_drop(&ast);
-}
-
-// static void dump_sema(const char *src, const char *fname) {
-//     spl_tok_vec_t toks = spl_lex(src, fname);
-//     spl_ast_t ast;
-//     spl_ast_init(&ast, &toks);
-//     spl_ast_prase(&ast);
-//     if (ast.parsed < 0) {
-//         printf("parse failed, skip sema dump\n");
-//         spl_ast_drop(&ast);
-//         return;
-//     }
-//     spl_ast_valid(&ast);
-//     spl_sema_t sema;
-//     spl_sema_init(&sema);
-//     sema.ast = &ast;
-//     spl_sema_run(&sema);
-//     spl_sema_check(&sema);
-//     printf("Sema  root_scope=%zu  scopes=%zu  errors=%d\n", sema.root_scope, sema.scopes.size,
-//            sema.error_count);
-//     for (usize i = 0; i < sema.scopes.size; i++) {
-//         printf("scope[%zu] parent=%zu\n", i, sema.scopes.data[i].parent);
-//         map_for(sema.scopes.data[i].symbols, mi) {
-//             printf("  %s -> def#%zu\n", sema.scopes.data[i].symbols.data[mi].key,
-//                    sema.scopes.data[i].symbols.data[mi].val);
-//         }
-//     }
-//     printf("TypeTable:\n");
-//     for (usize i = 0; i < sema.type.type_table.size; i++) {
-//         printf("  id#%zu type=", i);
-//         spl_type_pure_dump(&sema.type, i);
-//         printf("\n");
-//     }
-//     printf("DefTable:\n");
-//     for (usize i = 0; i < sema.type.def_table.size; i++) {
-//         printf("  def#%zu ", i);
-//         spl_type_def_dump(&sema.type, i);
-//         printf("\n");
-//     }
-//     spl_sema_drop(&sema);
-//     spl_ast_drop(&ast);
-// }
-
-// static void dump_ir(const char *src, const char *fname) {
-//     spl_tok_vec_t toks = spl_lex(src, fname);
-//     spl_ast_t ast;
-//     spl_ast_init(&ast, &toks);
-//     spl_ast_prase(&ast);
-//     if (ast.parsed < 0) {
-//         printf("parse failed, skip IR\n");
-//         spl_ast_drop(&ast);
-//         return;
-//     }
-//     spl_ast_valid(&ast);
-//     spl_sema_t sema;
-//     spl_sema_init(&sema);
-//     sema.ast = &ast;
-//     spl_sema_run(&sema);
-//     spl_sema_check(&sema);
-//     if (sema.error_count) {
-//         printf("sema errors=%d, skip IR\n", sema.error_count);
-//         spl_sema_drop(&sema);
-//         spl_ast_drop(&ast);
-//         return;
-//     }
-//     spl_ast2ir_t a2ir;
-//     spl_ast2ir_init(&a2ir, &sema);
-//     spl_ast2ir_run(&a2ir);
-//     if (a2ir.err_count)
-//         printf("ast2ir errors=%d\n", a2ir.err_count);
-//     spl_ir_dump(&a2ir.ir, &sema.type);
-//     spl_ast2ir_drop(&a2ir);
-//     spl_sema_drop(&sema);
-//     spl_ast_drop(&ast);
-// }
-
-static int cmd_dump(const char *flags, const char *path) {
-    long len;
-    char *src = read_file(path, &len);
-    if (!src)
+        printf("parse failed, no output\n");
+        // memory leak
         return 1;
-    int do_tokens = strstr(flags, "tokens") != NULL || strcmp(flags, "all") == 0;
-    int do_ast = strstr(flags, "ast") != NULL || strcmp(flags, "all") == 0;
-    int do_sema = strstr(flags, "sema") != NULL || strcmp(flags, "all") == 0;
-    int do_ir = strstr(flags, "ir") != NULL || strcmp(flags, "all") == 0;
-    if (do_tokens)
-        dump_tokens(src, path);
-    if (do_ast)
-        dump_ast(src, path);
-    // if (do_sema)
-    //     dump_sema(src, path);
-    // if (do_ir)
-    //     dump_ir(src, path);
-    free(src);
-    return 0;
-}
+    }
+    // spl_ast_valid(&ast); TODO
+    if (dump == DUMP_AST) {
+        spl_ast_dump(&ast, ast.root);
+        // memory leak
+        return 0;
+    }
+    spl_sema_t sema;
+    spl_type_t type;
+    spl_scope_t scope;
+    spl_type_init(&type);
+    spl_scope_init(&scope);
+    spl_sema_init(&sema, &ast, &type, &scope);
+    sema.ast = &ast;
+    spl_sema_run(&sema);
+    if (sema.error_count) {
+        printf("sema errors=%d, no output\n", sema.error_count);
+        // memory leak
+        return 1;
+    }
+    if (dump == DUMP_SEMA) {
+        spl_sema_dump(&sema);
+        spl_ast_dump(&ast, ast.root);
+        // memory leak
+        return 0;
+    }
 
-/* -g：在 .sir 尾部追加 debug 段（文本：IR 行 + VAR 行），spl_cli -g 读取 */
-static void gen_debug_map(const char *outpath, spl_ast_t *ast, const spl_ir_t *ir,
-                          const spl_ir2vm_t *ir2vm) {
-    FILE *f = fopen(outpath, "ab");
-    if (!f)
-        return;
-    fprintf(f, "SPLDBG\n");
-    // for (usize i = 0; i < ir2vm->fdbg.size; i++) {
-    //     const spl_ir2vm_fdbg_t *fd = &ir2vm->fdbg.data[i];
-    //     if (fd->fid >= ir->funcs.size)
-    //         continue;
-    //     const spl_ir_func_t *fn = &ir->funcs.data[fd->fid];
-    //     const char *fname = fn->name ? fn->name : "?";
-    //     for (usize ref = 1; ref < fd->node_first_ip.size; ref++) {
-    //         usize ip = fd->node_first_ip.data[ref];
-    //         if (ip == (usize)-1)
-    //             continue;
-    //         const spl_ir_node_t *n = &fn->nodes.data[ref];
-    //         fprintf(f, "IR %zu %zu %d %s\n", ip, ref, ast_line(ast, n->src_ref),
-    //                 spl_ir_kind_name(n->kind));
-    //     }
-    //     for (usize j = 0; j < fn->dbg_vars.size; j++) {
-    //         const spl_ir_dbg_var_t *dv = &fn->dbg_vars.data[j];
-    //         if (!dv->name)
-    //             continue;
-    //         fprintf(f, "VAR %s %s %zu %zu %d\n", fname, dv->name, dv->offset, dv->tid,
-    //                 dv->is_param);
-    //     }
-    // }
-    fclose(f);
-}
+    spl_ast2ir_t a2ir;
+    spl_ir_builder_t ir_builder;
+    spl_ir_builder_init(&ir_builder);
+    spl_ast2ir_init(&a2ir, &ir_builder, &sema);
+    spl_ast2ir_run(&a2ir);
+    if (a2ir.err_count) {
+        printf("ast2ir errors=%d, no output\n", a2ir.err_count);
+        spl_ast2ir_drop(&a2ir);
+        spl_sema_drop(&sema);
+        spl_ast_drop(&ast);
+        return 1;
+    }
+    if (dump == DUMP_IR) {
+        spl_ir_dump(&ir_builder.ir, &type);
+        // memory leak
+        return 0;
+    }
 
-static int compile_spl(const char *src, const char *fname, const char *outpath, int gen_debug) {
-    // spl_tok_vec_t toks = spl_lex(src, fname);
-    // spl_ast_t ast;
-    // spl_ast_init(&ast, &toks);
-    // spl_ast_prase(&ast);
-    // if (ast.parsed < 0) {
-    //     printf("parse failed, no output\n");
-    //     spl_ast_drop(&ast);
-    //     return 1;
-    // }
-    // spl_ast_valid(&ast);
-    // spl_sema_t sema;
-    // spl_sema_init(&sema);
-    // sema.ast = &ast;
-    // spl_sema_run(&sema);
-    // spl_sema_check(&sema);
-    // if (sema.error_count) {
-    //     printf("sema errors=%d, no output\n", sema.error_count);
-    //     spl_sema_drop(&sema);
-    //     spl_ast_drop(&ast);
-    //     return 1;
-    // }
-    // spl_ast2ir_t a2ir;
-    // spl_ast2ir_init(&a2ir, &sema);
-    // spl_ast2ir_run(&a2ir);
-    // if (a2ir.err_count) {
-    //     printf("ast2ir errors=%d, no output\n", a2ir.err_count);
-    //     spl_ast2ir_drop(&a2ir);
-    //     spl_sema_drop(&sema);
-    //     spl_ast_drop(&ast);
-    //     return 1;
-    // }
     // spl_ir2vm_t ir2vm;
     // spl_ir2vm_init(&ir2vm, &a2ir.ir, &sema.type);
     // int rc = spl_ir2vm_run(&ir2vm, outpath);
@@ -252,7 +146,7 @@ int main(int argc, char **argv) {
     }
     if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-h") == 0) {
         LOG_INFO("splc0 <in> <out>              compile (.spl -> .sir, stage B)");
-        LOG_INFO("splc0 --dump <flags> <file>   dump: tokens,ast,sema,ir,all");
+        LOG_INFO("splc0 --dump <flags> <file>   dump: lex,ast,sema,ir");
         return 0;
     }
     int argi = 1;
@@ -260,12 +154,28 @@ int main(int argc, char **argv) {
         LOG_FATAL("Usage: splc0 [--dump <flags>] <in> [out]");
         return 1;
     }
+
+    dump_t dump = DUMP_NONE;
     if (strcmp(argv[argi], "--dump") == 0) {
         if (argc < argi + 3) {
             LOG_INFO("splc0: --dump need <flags> <file>");
             return 1;
         }
-        return cmd_dump(argv[argi + 1], argv[argi + 2]);
+
+        const char *flags = argv[argi + 1];
+        if (strstr(flags, "lex") != NULL) {
+            dump = DUMP_TOKEN;
+        }
+        if (strstr(flags, "ast") != NULL) {
+            dump = DUMP_AST;
+        }
+        if (strstr(flags, "sema") != NULL) {
+            dump = DUMP_SEMA;
+        }
+        if (strstr(flags, "ir") != NULL) {
+            dump = DUMP_IR;
+        }
+        argi += 2;
     }
     /* splc0 [-g] <in> <out> */
     if (argc < argi + 2) {
@@ -285,7 +195,7 @@ int main(int argc, char **argv) {
     char *src = read_file(argv[argi], &len);
     if (!src)
         return 1;
-    int rc = compile_spl(src, argv[argi], argv[argi + 1], gen_debug);
+    int rc = compile_spl(src, argv[argi], argv[argi + 1], gen_debug, dump);
     free(src);
     return rc;
 }