cached bootstrap

internal/bootstrap.cm

@@ -9,11 +9,31 @@ function use_embed(name) {
 
 var fd = use_embed('fd')
 var json = use_embed('json')
+var crypto = use_embed('crypto')
 
 var use_cache = {}
 use_cache['fd'] = fd
 use_cache['os'] = os
 use_cache['json'] = json
+use_cache['crypto'] = crypto
+
+function content_hash(content) {
+  return text(crypto.blake2(content), 'h')
+}
+
+function cache_path(hash) {
+  if (!shop_path) return null
+  return shop_path + '/build/' + hash + '.mach'
+}
+
+function ensure_build_dir() {
+  if (!shop_path) return null
+  var dir = shop_path + '/build'
+  if (!fd.is_dir(dir)) {
+    fd.mkdir(dir)
+  }
+  return dir
+}
 
 // Bootstrap: load tokenize.cm, parse.cm, fold.cm from pre-compiled mach bytecode
 function use_basic(path) {
@@ -24,19 +44,28 @@ function use_basic(path) {
   return result
 }
 
-// Load a module from .mach/.mcode bytecode (bootstrap modules have no source fallback)
+// Load a module from cached .mach or .mcode bytecode
 function boot_load(name, env) {
-  var mach_path = core_path + '/' + name + ".cm.mach"
-  var mcode_path = core_path + '/' + name + ".cm.mcode"
-  var data = null
+  var mcode_path = core_path + '/boot/' + name + ".cm.mcode"
+  var mcode_blob = null
+  var hash = null
+  var cached = null
   var mcode_json = null
-  if (fd.is_file(mach_path)) {
-    data = fd.slurp(mach_path)
-    return mach_load(data, env)
-  }
+  var mach_blob = null
   if (fd.is_file(mcode_path)) {
-    mcode_json = text(fd.slurp(mcode_path))
-    return mach_eval_mcode(name, mcode_json, env)
+    mcode_blob = fd.slurp(mcode_path)
+    hash = content_hash(mcode_blob)
+    cached = cache_path(hash)
+    if (cached && fd.is_file(cached)) {
+      return mach_load(fd.slurp(cached), env)
+    }
+    mcode_json = text(mcode_blob)
+    mach_blob = mach_compile_mcode_bin(name, mcode_json)
+    if (cached) {
+      ensure_build_dir()
+      fd.slurpwrite(cached, mach_blob)
+    }
+    return mach_load(mach_blob, env)
   }
   print("error: missing bootstrap bytecode: " + name + "\n")
   disrupt
@@ -58,43 +87,33 @@ var streamline_mod = null
 // Warn if any .cm source is newer than its compiled bytecode
 function check_mach_stale() {
   var sources = [
-    "tokenize.cm",
-    "parse.cm",
-    "fold.cm",
-    "mcode.cm",
-    "streamline.cm",
-    "qbe.cm",
-    "qbe_emit.cm",
-    "verify_ir.cm",
-    "internal/bootstrap.cm",
-    "internal/engine.cm"
+    {src: "tokenize.cm", mcode: "boot/tokenize.cm.mcode"},
+    {src: "parse.cm", mcode: "boot/parse.cm.mcode"},
+    {src: "fold.cm", mcode: "boot/fold.cm.mcode"},
+    {src: "mcode.cm", mcode: "boot/mcode.cm.mcode"},
+    {src: "streamline.cm", mcode: "boot/streamline.cm.mcode"},
+    {src: "qbe.cm", mcode: "boot/qbe.cm.mcode"},
+    {src: "qbe_emit.cm", mcode: "boot/qbe_emit.cm.mcode"},
+    {src: "verify_ir.cm", mcode: "boot/verify_ir.cm.mcode"},
+    {src: "internal/bootstrap.cm", mcode: "boot/bootstrap.cm.mcode"},
+    {src: "internal/engine.cm", mcode: "boot/engine.cm.mcode"}
   ]
   var stale = []
   var _i = 0
   var cm_path = null
-  var mach_path = null
   var mcode_path = null
   var cm_stat = null
   var compiled_stat = null
-  var best_mtime = null
+  var entry = null
   while (_i < length(sources)) {
-    cm_path = core_path + '/' + sources[_i]
-    mach_path = cm_path + '.mach'
-    mcode_path = cm_path + '.mcode'
-    best_mtime = null
-    if (fd.is_file(mach_path)) {
-      best_mtime = fd.stat(mach_path).mtime
-    }
-    if (fd.is_file(mcode_path)) {
+    entry = sources[_i]
+    cm_path = core_path + '/' + entry.src
+    mcode_path = core_path + '/' + entry.mcode
+    if (fd.is_file(mcode_path) && fd.is_file(cm_path)) {
       compiled_stat = fd.stat(mcode_path)
-      if (best_mtime == null || compiled_stat.mtime > best_mtime) {
-        best_mtime = compiled_stat.mtime
-      }
-    }
-    if (best_mtime != null && fd.is_file(cm_path)) {
       cm_stat = fd.stat(cm_path)
-      if (cm_stat.mtime > best_mtime) {
-        push(stale, sources[_i])
+      if (cm_stat.mtime > compiled_stat.mtime) {
+        push(stale, entry.src)
       }
     }
     _i = _i + 1
@@ -175,23 +194,33 @@ function run_ast_noopt(name, ast, env) {
 
 // Helper to load engine.cm and run it with given env
 function load_engine(env) {
-  var engine_path = core_path + '/internal/engine.cm.mach'
-  var mcode_path = core_path + '/internal/engine.cm.mcode'
-  var data = null
+  var mcode_path = core_path + '/boot/engine.cm.mcode'
+  var mcode_blob = null
+  var hash = null
+  var cached = null
   var mcode_json = null
+  var mach_blob = null
   var engine_src = null
   var engine_ast = null
-  if (fd.is_file(engine_path)) {
-    data = fd.slurp(engine_path)
-    return mach_load(data, env)
-  }
   if (fd.is_file(mcode_path)) {
-    mcode_json = text(fd.slurp(mcode_path))
-    return mach_eval_mcode('engine', mcode_json, env)
+    mcode_blob = fd.slurp(mcode_path)
+    hash = content_hash(mcode_blob)
+    cached = cache_path(hash)
+    if (cached && fd.is_file(cached)) {
+      return mach_load(fd.slurp(cached), env)
+    }
+    mcode_json = text(mcode_blob)
+    mach_blob = mach_compile_mcode_bin('engine', mcode_json)
+    if (cached) {
+      ensure_build_dir()
+      fd.slurpwrite(cached, mach_blob)
+    }
+    return mach_load(mach_blob, env)
   }
-  engine_path = core_path + '/internal/engine.cm'
-  engine_src = text(fd.slurp(engine_path))
-  engine_ast = analyze(engine_src, engine_path)
+  // Fallback: compile from source
+  var engine_cm = core_path + '/internal/engine.cm'
+  engine_src = text(fd.slurp(engine_cm))
+  engine_ast = analyze(engine_src, engine_cm)
   return run_ast('engine', engine_ast, env)
 }