bootsnap 1.4.8 → 1.6.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: bdadfdd9d316f3362a593314fc596293395f5cfdb2f4ea0985f8fc77c54a4c0e
-  data.tar.gz: 88e75f2440f3f7fdd3785437e9e3e4494e5bbe4c9e95437f81fde4ea4845b884
+  metadata.gz: '029d63ba428f470d2cd2ef194d857e20689e7c8e377e2eaaaab83570f366034a'
+  data.tar.gz: 36a55f9d12dddef6ea3c4739f586c9236d7f4bc677a8b6747dfd7465d46eeca2
 SHA512:
-  metadata.gz: 31d24bdddce9bdb204f7fa346c7089bccac4b11081b8f172515883fc75fc77c1a2becab479753d7ebda52050ee3bd2790c54b3e32fd6100b07d09fa849d07134
-  data.tar.gz: 180abf489969e8629dec58d631553287f8f6546efcadb87f9e5caaf286bfc2e9d5c64f90f617e2626640862d5a31f347e1b69d9a225f9b3fdcc279bdb9cf7d66
+  metadata.gz: 131ec17c4e4912f387c18778250e689467ebfcc80e91eb884237307af1a57a3c89d4fb20c772fbc0330123a796d631861a9cf145bd32c54183077bbc97d7fa6f
+  data.tar.gz: d0c92454c8a5d8b16a25908fd0ae134fc817c5977958bde8424baca2bb6c96e60eb648c9f770bf7c7f58a3dd98871d4e7b0e3a0950c269100fded45ca9fddfa3

data/CHANGELOG.md

@@ -1,3 +1,26 @@
+# Unreleased
+
+# 1.6.0
+
+* Fix a Ruby 2.7/3.0 issue with `YAML.load_file` keyword arguments. (#342)
+* `bootsnap precompile` CLI use multiple processes to complete faster. (#341)
+* `bootsnap precompile` CLI also precompile YAML files. (#340)
+* Changed the load path cache directory from `$BOOTSNAP_CACHE_DIR/bootsnap-load-path-cache` to `$BOOTSNAP_CACHE_DIR/bootsnap/load-path-cache` for ease of use. (#334)
+* Changed the compile cache directory from `$BOOTSNAP_CACHE_DIR/bootsnap-compile-cache` to `$BOOTSNAP_CACHE_DIR/bootsnap/compile-cache` for ease of use. (#334)
+
+# 1.5.1
+
+* Workaround a Ruby bug in InstructionSequence.compile_file. (#332)
+
+# 1.5.0
+
+* Add a command line to statically precompile the ISeq cache. (#326)
+
+# 1.4.9
+
+* [Windows support](https://github.com/Shopify/bootsnap/pull/319)
+* [Fix potential crash](https://github.com/Shopify/bootsnap/pull/322)
+
 # 1.4.8
 
 * [Prevent FallbackScan from polluting exception cause](https://github.com/Shopify/bootsnap/pull/314)

data/README.md

@@ -1,4 +1,4 @@
-# Bootsnap [![Build Status](https://travis-ci.org/Shopify/bootsnap.svg?branch=master)](https://travis-ci.org/Shopify/bootsnap)
+# Bootsnap [![Actions Status](https://github.com/Shopify/bootsnap/workflows/ci/badge.svg)](https://github.com/Shopify/bootsnap/actions)
 
 Bootsnap is a library that plugs into Ruby, with optional support for `ActiveSupport` and `YAML`,
 to optimize and cache expensive computations. See [How Does This Work](#how-does-this-work).
@@ -29,7 +29,8 @@ If you are using Rails, add this to `config/boot.rb` immediately after `require
 require 'bootsnap/setup'
 ```
 
-Note that bootsnap writes to `tmp/cache`, and that directory *must* be writable. Rails will fail to
+Note that bootsnap writes to `tmp/cache` (or the path specified by `ENV['BOOTSNAP_CACHE_DIR']`), 
+and that directory *must* be writable. Rails will fail to
 boot if it is not. If this is unacceptable (e.g. you are running in a read-only container and
 unwilling to mount in a writable tmpdir), you should remove this line or wrap it in a conditional.
 
@@ -294,6 +295,19 @@ open    /c/nope.bundle -> -1
 # (nothing!)
 ```
 
+## Precompilation
+
+In development environments the bootsnap compilation cache is generated on the fly when source files are loaded.
+But in production environments, such as docker images, you might need to precompile the cache.
+
+To do so you can use the `bootsnap precompile` command.
+
+Example:
+
+```bash
+$ bundle exec bootsnap precompile --gemfile app/ lib/
+```
+
 ## When not to use Bootsnap
 
 *Alternative engines*: Bootsnap is pretty reliant on MRI features, and parts are disabled entirely on alternative ruby

data/exe/bootsnap

@@ -0,0 +1,5 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+require 'bootsnap/cli'
+exit Bootsnap::CLI.new(ARGV).run

data/ext/bootsnap/bootsnap.c

@@ -70,7 +70,7 @@ struct bs_cache_key {
 STATIC_ASSERT(sizeof(struct bs_cache_key) == KEY_SIZE);
 
 /* Effectively a schema version. Bumping invalidates all previous caches */
-static const uint32_t current_version = 2;
+static const uint32_t current_version = 3;
 
 /* hash of e.g. "x86_64-darwin17", invalidating when ruby is recompiled on a
  * new OS ABI, etc. */
@@ -91,16 +91,18 @@ static ID uncompilable;
 
 /* Functions exposed as module functions on Bootsnap::CompileCache::Native */
 static VALUE bs_compile_option_crc32_set(VALUE self, VALUE crc32_v);
-static VALUE bs_rb_fetch(VALUE self, VALUE cachedir_v, VALUE path_v, VALUE handler);
+static VALUE bs_rb_fetch(VALUE self, VALUE cachedir_v, VALUE path_v, VALUE handler, VALUE args);
+static VALUE bs_rb_precompile(VALUE self, VALUE cachedir_v, VALUE path_v, VALUE handler);
 
 /* Helpers */
 static uint64_t fnv1a_64(const char *str);
 static void bs_cache_path(const char * cachedir, const char * path, char (* cache_path)[MAX_CACHEPATH_SIZE]);
 static int bs_read_key(int fd, struct bs_cache_key * key);
 static int cache_key_equal(struct bs_cache_key * k1, struct bs_cache_key * k2);
-static VALUE bs_fetch(char * path, VALUE path_v, char * cache_path, VALUE handler);
+static VALUE bs_fetch(char * path, VALUE path_v, char * cache_path, VALUE handler, VALUE args);
+static VALUE bs_precompile(char * path, VALUE path_v, char * cache_path, VALUE handler);
 static int open_current_file(char * path, struct bs_cache_key * key, const char ** errno_provenance);
-static int fetch_cached_data(int fd, ssize_t data_size, VALUE handler, VALUE * output_data, int * exception_tag, const char ** errno_provenance);
+static int fetch_cached_data(int fd, ssize_t data_size, VALUE handler, VALUE args, VALUE * output_data, int * exception_tag, const char ** errno_provenance);
 static uint32_t get_ruby_revision(void);
 static uint32_t get_ruby_platform(void);
 
@@ -108,12 +110,12 @@ static uint32_t get_ruby_platform(void);
  * Helper functions to call ruby methods on handler object without crashing on
  * exception.
  */
-static int bs_storage_to_output(VALUE handler, VALUE storage_data, VALUE * output_data);
+static int bs_storage_to_output(VALUE handler, VALUE args, VALUE storage_data, VALUE * output_data);
 static VALUE prot_storage_to_output(VALUE arg);
 static VALUE prot_input_to_output(VALUE arg);
-static void bs_input_to_output(VALUE handler, VALUE input_data, VALUE * output_data, int * exception_tag);
+static void bs_input_to_output(VALUE handler, VALUE args, VALUE input_data, VALUE * output_data, int * exception_tag);
 static VALUE prot_input_to_storage(VALUE arg);
-static int bs_input_to_storage(VALUE handler, VALUE input_data, VALUE pathval, VALUE * storage_data);
+static int bs_input_to_storage(VALUE handler, VALUE args, VALUE input_data, VALUE pathval, VALUE * storage_data);
 struct s2o_data;
 struct i2o_data;
 struct i2s_data;
@@ -148,7 +150,8 @@ Init_bootsnap(void)
   uncompilable = rb_intern("__bootsnap_uncompilable__");
 
   rb_define_module_function(rb_mBootsnap_CompileCache_Native, "coverage_running?", bs_rb_coverage_running, 0);
-  rb_define_module_function(rb_mBootsnap_CompileCache_Native, "fetch", bs_rb_fetch, 3);
+  rb_define_module_function(rb_mBootsnap_CompileCache_Native, "fetch", bs_rb_fetch, 4);
+  rb_define_module_function(rb_mBootsnap_CompileCache_Native, "precompile", bs_rb_precompile, 3);
   rb_define_module_function(rb_mBootsnap_CompileCache_Native, "compile_option_crc32=", bs_compile_option_crc32_set, 1);
 
   current_umask = umask(0777);
@@ -267,7 +270,6 @@ static void
 bs_cache_path(const char * cachedir, const char * path, char (* cache_path)[MAX_CACHEPATH_SIZE])
 {
   uint64_t hash = fnv1a_64(path);
-
   uint8_t first_byte = (hash >> (64 - 8));
   uint64_t remainder = hash & 0x00ffffffffffffff;
 
@@ -301,7 +303,7 @@ cache_key_equal(struct bs_cache_key * k1, struct bs_cache_key * k2)
  * conversions on the ruby VALUE arguments before passing them along.
  */
 static VALUE
-bs_rb_fetch(VALUE self, VALUE cachedir_v, VALUE path_v, VALUE handler)
+bs_rb_fetch(VALUE self, VALUE cachedir_v, VALUE path_v, VALUE handler, VALUE args)
 {
   FilePathValue(path_v);
 
@@ -319,9 +321,35 @@ bs_rb_fetch(VALUE self, VALUE cachedir_v, VALUE path_v, VALUE handler)
   /* generate cache path to cache_path */
   bs_cache_path(cachedir, path, &cache_path);
 
-  return bs_fetch(path, path_v, cache_path, handler);
+  return bs_fetch(path, path_v, cache_path, handler, args);
 }
 
+/*
+ * Entrypoint for Bootsnap::CompileCache::Native.precompile.
+ * Similar to fetch, but it only generate the cache if missing
+ * and doesn't return the content.
+ */
+static VALUE
+bs_rb_precompile(VALUE self, VALUE cachedir_v, VALUE path_v, VALUE handler)
+{
+  FilePathValue(path_v);
+
+  Check_Type(cachedir_v, T_STRING);
+  Check_Type(path_v, T_STRING);
+
+  if (RSTRING_LEN(cachedir_v) > MAX_CACHEDIR_SIZE) {
+    rb_raise(rb_eArgError, "cachedir too long");
+  }
+
+  char * cachedir = RSTRING_PTR(cachedir_v);
+  char * path     = RSTRING_PTR(path_v);
+  char cache_path[MAX_CACHEPATH_SIZE];
+
+  /* generate cache path to cache_path */
+  bs_cache_path(cachedir, path, &cache_path);
+
+  return bs_precompile(path, path_v, cache_path, handler);
+}
 /*
  * Open the file we want to load/cache and generate a cache key for it if it
  * was loaded.
@@ -428,7 +456,7 @@ open_cache_file(const char * path, struct bs_cache_key * key, const char ** errn
  * or exception, will be the final data returnable to the user.
  */
 static int
-fetch_cached_data(int fd, ssize_t data_size, VALUE handler, VALUE * output_data, int * exception_tag, const char ** errno_provenance)
+fetch_cached_data(int fd, ssize_t data_size, VALUE handler, VALUE args, VALUE * output_data, int * exception_tag, const char ** errno_provenance)
 {
   char * data = NULL;
   ssize_t nread;
@@ -454,9 +482,9 @@ fetch_cached_data(int fd, ssize_t data_size, VALUE handler, VALUE * output_data,
     goto done;
   }
 
-  storage_data = rb_str_new_static(data, data_size);
+  storage_data = rb_str_new(data, data_size);
 
-  *exception_tag = bs_storage_to_output(handler, storage_data, output_data);
+  *exception_tag = bs_storage_to_output(handler, args, storage_data, output_data);
   ret = 0;
 done:
   if (data != NULL) xfree(data);
@@ -624,7 +652,7 @@ bs_read_contents(int fd, size_t size, char ** contents, const char ** errno_prov
  *   - Return storage_to_output(storage_data)
  */
 static VALUE
-bs_fetch(char * path, VALUE path_v, char * cache_path, VALUE handler)
+bs_fetch(char * path, VALUE path_v, char * cache_path, VALUE handler, VALUE args)
 {
   struct bs_cache_key cached_key, current_key;
   char * contents = NULL;
@@ -657,7 +685,7 @@ bs_fetch(char * path, VALUE path_v, char * cache_path, VALUE handler)
   if (valid_cache) {
     /* Fetch the cache data and return it if we're able to load it successfully */
     res = fetch_cached_data(
-      cache_fd, (ssize_t)cached_key.data_size, handler,
+      cache_fd, (ssize_t)cached_key.data_size, handler, args,
       &output_data, &exception_tag, &errno_provenance
     );
     if (exception_tag != 0)                   goto raise;
@@ -671,15 +699,15 @@ bs_fetch(char * path, VALUE path_v, char * cache_path, VALUE handler)
 
   /* Read the contents of the source file into a buffer */
   if (bs_read_contents(current_fd, current_key.size, &contents, &errno_provenance) < 0) goto fail_errno;
-  input_data = rb_str_new_static(contents, current_key.size);
+  input_data = rb_str_new(contents, current_key.size);
 
   /* Try to compile the input_data using input_to_storage(input_data) */
-  exception_tag = bs_input_to_storage(handler, input_data, path_v, &storage_data);
+  exception_tag = bs_input_to_storage(handler, args, input_data, path_v, &storage_data);
   if (exception_tag != 0) goto raise;
   /* If input_to_storage raised Bootsnap::CompileCache::Uncompilable, don't try
    * to cache anything; just return input_to_output(input_data) */
   if (storage_data == uncompilable) {
-    bs_input_to_output(handler, input_data, &output_data, &exception_tag);
+    bs_input_to_output(handler, args, input_data, &output_data, &exception_tag);
     if (exception_tag != 0) goto raise;
     goto succeed;
   }
@@ -691,7 +719,7 @@ bs_fetch(char * path, VALUE path_v, char * cache_path, VALUE handler)
   if (res < 0) goto fail_errno;
 
   /* Having written the cache, now convert storage_data to output_data */
-  exception_tag = bs_storage_to_output(handler, storage_data, &output_data);
+  exception_tag = bs_storage_to_output(handler, args, storage_data, &output_data);
   if (exception_tag != 0) goto raise;
 
   /* If output_data is nil, delete the cache entry and generate the output
@@ -701,7 +729,7 @@ bs_fetch(char * path, VALUE path_v, char * cache_path, VALUE handler)
       errno_provenance = "bs_fetch:unlink";
       goto fail_errno;
     }
-    bs_input_to_output(handler, input_data, &output_data, &exception_tag);
+    bs_input_to_output(handler, args, input_data, &output_data, &exception_tag);
     if (exception_tag != 0) goto raise;
   }
 
@@ -732,6 +760,79 @@ invalid_type_storage_data:
 #undef CLEANUP
 }
 
+static VALUE
+bs_precompile(char * path, VALUE path_v, char * cache_path, VALUE handler)
+{
+  struct bs_cache_key cached_key, current_key;
+  char * contents = NULL;
+  int cache_fd = -1, current_fd = -1;
+  int res, valid_cache = 0, exception_tag = 0;
+  const char * errno_provenance = NULL;
+
+  VALUE input_data;   /* data read from source file, e.g. YAML or ruby source */
+  VALUE storage_data; /* compiled data, e.g. msgpack / binary iseq */
+
+  /* Open the source file and generate a cache key for it */
+  current_fd = open_current_file(path, &current_key, &errno_provenance);
+  if (current_fd < 0) goto fail;
+
+  /* Open the cache key if it exists, and read its cache key in */
+  cache_fd = open_cache_file(cache_path, &cached_key, &errno_provenance);
+  if (cache_fd == CACHE_MISSING_OR_INVALID) {
+    /* This is ok: valid_cache remains false, we re-populate it. */
+  } else if (cache_fd < 0) {
+    goto fail;
+  } else {
+    /* True if the cache existed and no invalidating changes have occurred since
+     * it was generated. */
+    valid_cache = cache_key_equal(&current_key, &cached_key);
+  }
+
+  if (valid_cache) {
+    goto succeed;
+  }
+
+  close(cache_fd);
+  cache_fd = -1;
+  /* Cache is stale, invalid, or missing. Regenerate and write it out. */
+
+  /* Read the contents of the source file into a buffer */
+  if (bs_read_contents(current_fd, current_key.size, &contents, &errno_provenance) < 0) goto fail;
+  input_data = rb_str_new(contents, current_key.size);
+
+  /* Try to compile the input_data using input_to_storage(input_data) */
+  exception_tag = bs_input_to_storage(handler, Qnil, input_data, path_v, &storage_data);
+  if (exception_tag != 0) goto fail;
+
+  /* If input_to_storage raised Bootsnap::CompileCache::Uncompilable, don't try
+   * to cache anything; just return false */
+  if (storage_data == uncompilable) {
+    goto fail;
+  }
+  /* If storage_data isn't a string, we can't cache it */
+  if (!RB_TYPE_P(storage_data, T_STRING)) goto fail;
+
+  /* Write the cache key and storage_data to the cache directory */
+  res = atomic_write_cache_file(cache_path, &current_key, storage_data, &errno_provenance);
+  if (res < 0) goto fail;
+
+  goto succeed;
+
+#define CLEANUP \
+  if (contents != NULL) xfree(contents);   \
+  if (current_fd >= 0)  close(current_fd); \
+  if (cache_fd >= 0)    close(cache_fd);
+
+succeed:
+  CLEANUP;
+  return Qtrue;
+fail:
+  CLEANUP;
+  return Qfalse;
+#undef CLEANUP
+}
+
+
 /*****************************************************************************/
 /********************* Handler Wrappers **************************************/
 /*****************************************************************************
@@ -751,11 +852,13 @@ invalid_type_storage_data:
 
 struct s2o_data {
   VALUE handler;
+  VALUE args;
   VALUE storage_data;
 };
 
 struct i2o_data {
   VALUE handler;
+  VALUE args;
   VALUE input_data;
 };
 
@@ -769,15 +872,16 @@ static VALUE
 prot_storage_to_output(VALUE arg)
 {
   struct s2o_data * data = (struct s2o_data *)arg;
-  return rb_funcall(data->handler, rb_intern("storage_to_output"), 1, data->storage_data);
+  return rb_funcall(data->handler, rb_intern("storage_to_output"), 2, data->storage_data, data->args);
 }
 
 static int
-bs_storage_to_output(VALUE handler, VALUE storage_data, VALUE * output_data)
+bs_storage_to_output(VALUE handler, VALUE args, VALUE storage_data, VALUE * output_data)
 {
   int state;
   struct s2o_data s2o_data = {
     .handler      = handler,
+    .args         = args,
     .storage_data = storage_data,
   };
   *output_data = rb_protect(prot_storage_to_output, (VALUE)&s2o_data, &state);
@@ -785,10 +889,11 @@ bs_storage_to_output(VALUE handler, VALUE storage_data, VALUE * output_data)
 }
 
 static void
-bs_input_to_output(VALUE handler, VALUE input_data, VALUE * output_data, int * exception_tag)
+bs_input_to_output(VALUE handler, VALUE args, VALUE input_data, VALUE * output_data, int * exception_tag)
 {
   struct i2o_data i2o_data = {
     .handler    = handler,
+    .args       = args,
     .input_data = input_data,
   };
   *output_data = rb_protect(prot_input_to_output, (VALUE)&i2o_data, exception_tag);
@@ -798,7 +903,7 @@ static VALUE
 prot_input_to_output(VALUE arg)
 {
   struct i2o_data * data = (struct i2o_data *)arg;
-  return rb_funcall(data->handler, rb_intern("input_to_output"), 1, data->input_data);
+  return rb_funcall(data->handler, rb_intern("input_to_output"), 2, data->input_data, data->args);
 }
 
 static VALUE
@@ -825,7 +930,7 @@ prot_input_to_storage(VALUE arg)
 }
 
 static int
-bs_input_to_storage(VALUE handler, VALUE input_data, VALUE pathval, VALUE * storage_data)
+bs_input_to_storage(VALUE handler, VALUE args, VALUE input_data, VALUE pathval, VALUE * storage_data)
 {
   int state;
   struct i2s_data i2s_data = {

data/lib/bootsnap.rb

@@ -24,13 +24,13 @@ module Bootsnap
     setup_disable_trace if disable_trace
 
     Bootsnap::LoadPathCache.setup(
-      cache_path:       cache_dir + '/bootsnap-load-path-cache',
+      cache_path:       cache_dir + '/bootsnap/load-path-cache',
       development_mode: development_mode,
       active_support:   autoload_paths_cache
     ) if load_path_cache
 
     Bootsnap::CompileCache.setup(
-      cache_dir: cache_dir + '/bootsnap-compile-cache',
+      cache_dir: cache_dir + '/bootsnap/compile-cache',
       iseq: compile_cache_iseq,
       yaml: compile_cache_yaml
     )

data/lib/bootsnap/cli.rb

@@ -0,0 +1,246 @@
+# frozen_string_literal: true
+
+require 'bootsnap'
+require 'bootsnap/cli/worker_pool'
+require 'optparse'
+require 'fileutils'
+require 'etc'
+
+module Bootsnap
+  class CLI
+    unless Regexp.method_defined?(:match?)
+      module RegexpMatchBackport
+        refine Regexp do
+          def match?(string)
+            !!match(string)
+          end
+        end
+      end
+      using RegexpMatchBackport
+    end
+
+    attr_reader :cache_dir, :argv
+
+    attr_accessor :compile_gemfile, :exclude, :verbose, :iseq, :yaml, :jobs
+
+    def initialize(argv)
+      @argv = argv
+      self.cache_dir = ENV.fetch('BOOTSNAP_CACHE_DIR', 'tmp/cache')
+      self.compile_gemfile = false
+      self.exclude = nil
+      self.verbose = false
+      self.jobs = Etc.nprocessors
+      self.iseq = true
+      self.yaml = true
+    end
+
+    def precompile_command(*sources)
+      require 'bootsnap/compile_cache/iseq'
+      require 'bootsnap/compile_cache/yaml'
+
+      fix_default_encoding do
+        Bootsnap::CompileCache::ISeq.cache_dir = self.cache_dir
+        Bootsnap::CompileCache::YAML.init!
+        Bootsnap::CompileCache::YAML.cache_dir = self.cache_dir
+
+        @work_pool = WorkerPool.create(size: jobs, jobs: {
+          ruby: method(:precompile_ruby),
+          yaml: method(:precompile_yaml),
+        })
+        @work_pool.spawn
+
+        main_sources = sources.map { |d| File.expand_path(d) }
+        precompile_ruby_files(main_sources)
+        precompile_yaml_files(main_sources)
+
+        if compile_gemfile
+          # Some gems embed their tests, they're very unlikely to be loaded, so not worth precompiling.
+          gem_exclude = Regexp.union([exclude, '/spec/', '/test/'].compact)
+          precompile_ruby_files($LOAD_PATH.map { |d| File.expand_path(d) }, exclude: gem_exclude)
+
+          # Gems that include YAML files usually don't put them in `lib/`.
+          # So we look at the gem root.
+          gem_pattern = %r{^#{Regexp.escape(Bundler.bundle_path.to_s)}/?(?:bundler/)?gem\/[^/]+}
+          gem_paths = $LOAD_PATH.map { |p| p[gem_pattern] }.compact.uniq
+          precompile_yaml_files(gem_paths, exclude: gem_exclude)
+        end
+
+        if exitstatus = @work_pool.shutdown
+          exit(exitstatus)
+        end
+      end
+      0
+    end
+
+    dir_sort = begin
+      Dir[__FILE__, sort: false]
+      true
+    rescue ArgumentError, TypeError
+      false
+    end
+
+    if dir_sort
+      def list_files(path, pattern)
+        if File.directory?(path)
+          Dir[File.join(path,  pattern), sort: false]
+        elsif File.exist?(path)
+          [path]
+        else
+          []
+        end
+      end
+    else
+      def list_files(path, pattern)
+        if File.directory?(path)
+          Dir[File.join(path,  pattern)]
+        elsif File.exist?(path)
+          [path]
+        else
+          []
+        end
+      end
+    end
+
+    def run
+      parser.parse!(argv)
+      command = argv.shift
+      method = "#{command}_command"
+      if respond_to?(method)
+        public_send(method, *argv)
+      else
+        invalid_usage!("Unknown command: #{command}")
+      end
+    end
+
+    private
+
+    def precompile_yaml_files(load_paths, exclude: self.exclude)
+      return unless yaml
+
+      load_paths.each do |path|
+        if !exclude || !exclude.match?(path)
+          list_files(path, '**/*.{yml,yaml}').each do |yaml_file|
+            # We ignore hidden files to not match the various .ci.yml files
+            if !yaml_file.include?('/.') && (!exclude || !exclude.match?(yaml_file))
+              @work_pool.push(:yaml, yaml_file)
+            end
+          end
+        end
+      end
+    end
+
+    def precompile_yaml(*yaml_files)
+      Array(yaml_files).each do |yaml_file|
+        if CompileCache::YAML.precompile(yaml_file, cache_dir: cache_dir)
+          STDERR.puts(yaml_file) if verbose
+        end
+      end
+    end
+
+    def precompile_ruby_files(load_paths, exclude: self.exclude)
+      return unless iseq
+
+      load_paths.each do |path|
+        if !exclude || !exclude.match?(path)
+          list_files(path, '**/*.rb').each do |ruby_file|
+            if !exclude || !exclude.match?(ruby_file)
+              @work_pool.push(:ruby, ruby_file)
+            end
+          end
+        end
+      end
+    end
+
+    def precompile_ruby(*ruby_files)
+      Array(ruby_files).each do |ruby_file|
+        if CompileCache::ISeq.precompile(ruby_file, cache_dir: cache_dir)
+          STDERR.puts(ruby_file) if verbose
+        end
+      end
+    end
+
+    def fix_default_encoding
+      if Encoding.default_external == Encoding::US_ASCII
+        Encoding.default_external = Encoding::UTF_8
+        begin
+          yield
+        ensure
+          Encoding.default_external = Encoding::US_ASCII
+        end
+      else
+        yield
+      end
+    end
+
+    def invalid_usage!(message)
+      STDERR.puts message
+      STDERR.puts
+      STDERR.puts parser
+      1
+    end
+
+    def cache_dir=(dir)
+      @cache_dir = File.expand_path(File.join(dir, 'bootsnap/compile-cache'))
+    end
+
+    def exclude_pattern(pattern)
+      (@exclude_patterns ||= []) << Regexp.new(pattern)
+      self.exclude = Regexp.union(@exclude_patterns)
+    end
+
+    def parser
+      @parser ||= OptionParser.new do |opts|
+        opts.banner = "Usage: bootsnap COMMAND [ARGS]"
+        opts.separator ""
+        opts.separator "GLOBAL OPTIONS"
+        opts.separator ""
+
+        help = <<~EOS
+          Path to the bootsnap cache directory. Defaults to tmp/cache
+        EOS
+        opts.on('--cache-dir DIR', help.strip) do |dir|
+          self.cache_dir = dir
+        end
+
+        help = <<~EOS
+          Print precompiled paths.
+        EOS
+        opts.on('--verbose', '-v', help.strip) do
+          self.verbose = true
+        end
+
+        help = <<~EOS
+          Number of workers to use. Default to number of processors, set to 0 to disable multi-processing.
+        EOS
+        opts.on('--jobs JOBS', '-j', help.strip) do |jobs|
+          self.jobs = Integer(jobs)
+        end
+
+        opts.separator ""
+        opts.separator "COMMANDS"
+        opts.separator ""
+        opts.separator "    precompile [DIRECTORIES...]: Precompile all .rb files in the passed directories"
+
+        help = <<~EOS
+          Precompile the gems in Gemfile
+        EOS
+        opts.on('--gemfile', help) { self.compile_gemfile = true }
+
+        help = <<~EOS
+          Path pattern to not precompile. e.g. --exclude 'aws-sdk|google-api'
+        EOS
+        opts.on('--exclude PATTERN', help) { |pattern| exclude_pattern(pattern) }
+
+        help = <<~EOS
+          Disable ISeq (.rb) precompilation.
+        EOS
+        opts.on('--no-iseq', help) { self.iseq = false }
+
+        help = <<~EOS
+          Disable YAML precompilation.
+        EOS
+        opts.on('--no-yaml', help) { self.yaml = false }
+      end
+    end
+  end
+end

data/lib/bootsnap/cli/worker_pool.rb

@@ -0,0 +1,131 @@
+# frozen_string_literal: true
+
+module Bootsnap
+  class CLI
+    class WorkerPool
+      class << self
+        def create(size:, jobs:)
+          if size > 0 && Process.respond_to?(:fork)
+            new(size: size, jobs: jobs)
+          else
+            Inline.new(jobs: jobs)
+          end
+        end
+      end
+
+      class Inline
+        def initialize(jobs: {})
+          @jobs = jobs
+        end
+
+        def push(job, *args)
+          @jobs.fetch(job).call(*args)
+          nil
+        end
+
+        def spawn
+          # noop
+        end
+
+        def shutdown
+          # noop
+        end
+      end
+
+      class Worker
+        attr_reader :to_io, :pid
+
+        def initialize(jobs)
+          @jobs = jobs
+          @pipe_out, @to_io = IO.pipe
+          @pid = nil
+        end
+
+        def write(message, block: true)
+          payload = Marshal.dump(message)
+          if block
+            to_io.write(payload)
+            true
+          else
+            to_io.write_nonblock(payload, exception: false) != :wait_writable
+          end
+        end
+
+        def close
+          to_io.close
+        end
+
+        def work_loop
+          loop do
+            job, *args = Marshal.load(@pipe_out)
+            return if job == :exit
+            @jobs.fetch(job).call(*args)
+          end
+        rescue IOError
+          nil
+        end
+
+        def spawn
+          @pid = Process.fork do
+            to_io.close
+            work_loop
+            exit!(0)
+          end
+          @pipe_out.close
+          true
+        end
+      end
+
+      def initialize(size:, jobs: {})
+        @size = size
+        @jobs = jobs
+        @queue = Queue.new
+        @pids = []
+      end
+
+      def spawn
+        @workers = @size.times.map { Worker.new(@jobs) }
+        @workers.each(&:spawn)
+        @dispatcher_thread = Thread.new { dispatch_loop }
+        @dispatcher_thread.abort_on_exception = true
+        true
+      end
+
+      def dispatch_loop
+        loop do
+          case job = @queue.pop
+          when nil
+            @workers.each do |worker|
+              worker.write([:exit])
+              worker.close
+            end
+            return true
+          else
+            unless @workers.sample.write(job, block: false)
+              free_worker.write(job)
+            end
+          end
+        end
+      end
+
+      def free_worker
+        IO.select(nil, @workers)[1].sample
+      end
+
+      def push(*args)
+        @queue.push(args)
+        nil
+      end
+
+      def shutdown
+        @queue.close
+        @dispatcher_thread.join
+        @workers.each do |worker|
+          _pid, status = Process.wait2(worker.pid)
+          return status.exitstatus unless status.success?
+        end
+        nil
+      end
+    end
+  end
+end

data/lib/bootsnap/compile_cache.rb

@@ -34,9 +34,9 @@ module Bootsnap
     end
 
     def self.supported?
-      # only enable on 'ruby' (MRI), POSIX (darwin, linux, *bsd), and >= 2.3.0
+      # only enable on 'ruby' (MRI), POSIX (darwin, linux, *bsd), Windows (RubyInstaller2) and >= 2.3.0
       RUBY_ENGINE == 'ruby' &&
-      RUBY_PLATFORM =~ /darwin|linux|bsd/ &&
+      RUBY_PLATFORM =~ /darwin|linux|bsd|mswin|mingw|cygwin/ &&
       Gem::Version.new(RUBY_VERSION) >= Gem::Version.new("2.3.0")
     end
   end

data/lib/bootsnap/compile_cache/iseq.rb

@@ -15,7 +15,7 @@ module Bootsnap
         raise(Uncompilable, 'syntax error')
       end
 
-      def self.storage_to_output(binary)
+      def self.storage_to_output(binary, _args)
         RubyVM::InstructionSequence.load_from_binary(binary)
       rescue RuntimeError => e
         if e.message == 'broken binary format'
@@ -26,7 +26,24 @@ module Bootsnap
         end
       end
 
-      def self.input_to_output(_)
+      def self.fetch(path, cache_dir: ISeq.cache_dir)
+        Bootsnap::CompileCache::Native.fetch(
+          cache_dir,
+          path.to_s,
+          Bootsnap::CompileCache::ISeq,
+          nil,
+        )
+      end
+
+      def self.precompile(path, cache_dir: ISeq.cache_dir)
+        Bootsnap::CompileCache::Native.precompile(
+          cache_dir,
+          path.to_s,
+          Bootsnap::CompileCache::ISeq,
+        )
+      end
+
+      def self.input_to_output(_data, _kwargs)
         nil # ruby handles this
       end
 
@@ -35,11 +52,7 @@ module Bootsnap
           # Having coverage enabled prevents iseq dumping/loading.
           return nil if defined?(Coverage) && Bootsnap::CompileCache::Native.coverage_running?
 
-          Bootsnap::CompileCache::Native.fetch(
-            Bootsnap::CompileCache::ISeq.cache_dir,
-            path.to_s,
-            Bootsnap::CompileCache::ISeq
-          )
+          Bootsnap::CompileCache::ISeq.fetch(path.to_s)
         rescue Errno::EACCES
           Bootsnap::CompileCache.permission_error(path)
         rescue RuntimeError => e
@@ -60,6 +73,7 @@ module Bootsnap
         crc = Zlib.crc32(option.inspect)
         Bootsnap::CompileCache::Native.compile_option_crc32 = crc
       end
+      compile_option_updated
 
       def self.install!(cache_dir)
         Bootsnap::CompileCache::ISeq.cache_dir = cache_dir

data/lib/bootsnap/compile_cache/yaml.rb

@@ -5,58 +5,107 @@ module Bootsnap
   module CompileCache
     module YAML
       class << self
-        attr_accessor(:msgpack_factory)
-      end
+        attr_accessor(:msgpack_factory, :cache_dir, :supported_options)
 
-      def self.input_to_storage(contents, _)
-        raise(Uncompilable) if contents.index("!ruby/object")
-        obj = ::YAML.load(contents)
-        msgpack_factory.packer.write(obj).to_s
-      rescue NoMethodError, RangeError
-        # if the object included things that we can't serialize, fall back to
-        # Marshal. It's a bit slower, but can encode anything yaml can.
-        # NoMethodError is unexpected types; RangeError is Bignums
-        Marshal.dump(obj)
-      end
+        def input_to_storage(contents, _)
+          raise(Uncompilable) if contents.index("!ruby/object")
+          obj = ::YAML.load(contents)
+          msgpack_factory.dump(obj)
+        rescue NoMethodError, RangeError
+          # The object included things that we can't serialize
+          raise(Uncompilable)
+        end
 
-      def self.storage_to_output(data)
-        # This could have a meaning in messagepack, and we're being a little lazy
-        # about it. -- but a leading 0x04 would indicate the contents of the YAML
-        # is a positive integer, which is rare, to say the least.
-        if data[0] == 0x04.chr && data[1] == 0x08.chr
-          Marshal.load(data)
-        else
-          msgpack_factory.unpacker.feed(data).read
+        def storage_to_output(data, kwargs)
+          if kwargs && kwargs.key?(:symbolize_names)
+            kwargs[:symbolize_keys] = kwargs.delete(:symbolize_names)
+          end
+          msgpack_factory.load(data, kwargs)
         end
-      end
 
-      def self.input_to_output(data)
-        ::YAML.load(data)
-      end
+        def input_to_output(data, kwargs)
+          ::YAML.load(data, **(kwargs || {}))
+        end
+
+        def precompile(path, cache_dir: YAML.cache_dir)
+          Bootsnap::CompileCache::Native.precompile(
+            cache_dir,
+            path.to_s,
+            Bootsnap::CompileCache::YAML,
+          )
+        end
+
+        def install!(cache_dir)
+          self.cache_dir = cache_dir
+          init!
+          ::YAML.singleton_class.prepend(Patch)
+        end
 
-      def self.install!(cache_dir)
-        require('yaml')
-        require('msgpack')
+        def init!
+          require('yaml')
+          require('msgpack')
+          require('date')
 
-        # MessagePack serializes symbols as strings by default.
-        # We want them to roundtrip cleanly, so we use a custom factory.
-        # see: https://github.com/msgpack/msgpack-ruby/pull/122
-        factory = MessagePack::Factory.new
-        factory.register_type(0x00, Symbol)
-        Bootsnap::CompileCache::YAML.msgpack_factory = factory
+          # MessagePack serializes symbols as strings by default.
+          # We want them to roundtrip cleanly, so we use a custom factory.
+          # see: https://github.com/msgpack/msgpack-ruby/pull/122
+          factory = MessagePack::Factory.new
+          factory.register_type(0x00, Symbol)
+          factory.register_type(
+            MessagePack::Timestamp::TYPE, # or just -1
+            Time,
+            packer: MessagePack::Time::Packer,
+            unpacker: MessagePack::Time::Unpacker
+          )
+
+          marshal_fallback = {
+            packer: ->(value) { Marshal.dump(value) },
+            unpacker: ->(payload) { Marshal.load(payload) },
+          }
+          {
+            Date => 0x01,
+            Regexp => 0x02,
+          }.each do |type, code|
+            factory.register_type(code, type, marshal_fallback)
+          end
+
+          self.msgpack_factory = factory
+
+          self.supported_options = []
+          params = ::YAML.method(:load).parameters
+          if params.include?([:key, :symbolize_names])
+            self.supported_options << :symbolize_names
+          end
+          if params.include?([:key, :freeze])
+            if factory.load(factory.dump('yaml'), freeze: true).frozen?
+              self.supported_options << :freeze
+            end
+          end
+          self.supported_options.freeze
+        end
+      end
+
+      module Patch
+        def load_file(path, *args)
+          return super if args.size > 1
+          if kwargs = args.first
+            return super unless kwargs.is_a?(Hash)
+            return super unless (kwargs.keys - ::Bootsnap::CompileCache::YAML.supported_options).empty?
+          end
 
-        klass = class << ::YAML; self; end
-        klass.send(:define_method, :load_file) do |path|
           begin
-            Bootsnap::CompileCache::Native.fetch(
-              cache_dir,
+            ::Bootsnap::CompileCache::Native.fetch(
+              Bootsnap::CompileCache::YAML.cache_dir,
               path,
-              Bootsnap::CompileCache::YAML
+              ::Bootsnap::CompileCache::YAML,
+              kwargs,
             )
           rescue Errno::EACCES
-            Bootsnap::CompileCache.permission_error(path)
+            ::Bootsnap::CompileCache.permission_error(path)
           end
         end
+
+        ruby2_keywords :load_file if respond_to?(:ruby2_keywords, true)
       end
     end
   end

data/lib/bootsnap/load_path_cache.rb

@@ -61,7 +61,7 @@ module Bootsnap
 
       def supported?
         RUBY_ENGINE == 'ruby' &&
-        RUBY_PLATFORM =~ /darwin|linux|bsd/
+        RUBY_PLATFORM =~ /darwin|linux|bsd|mswin|mingw|cygwin/
       end
     end
   end

data/lib/bootsnap/version.rb

@@ -1,4 +1,4 @@
 # frozen_string_literal: true
 module Bootsnap
-  VERSION = "1.4.8"
+  VERSION = "1.6.0"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: bootsnap
 version: !ruby/object:Gem::Version
-  version: 1.4.8
+  version: 1.6.0
 platform: ruby
 authors:
 - Burke Libbey
 autorequire: 
-bindir: bin
+bindir: exe
 cert_chain: []
-date: 2020-08-11 00:00:00.000000000 Z
+date: 2021-01-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -97,7 +97,8 @@ dependencies:
 description: Boot large ruby/rails apps faster
 email:
 - burke.libbey@shopify.com
-executables: []
+executables:
+- bootsnap
 extensions:
 - ext/bootsnap/extconf.rb
 extra_rdoc_files: []
@@ -105,11 +106,14 @@ files:
 - CHANGELOG.md
 - LICENSE.txt
 - README.md
+- exe/bootsnap
 - ext/bootsnap/bootsnap.c
 - ext/bootsnap/bootsnap.h
 - ext/bootsnap/extconf.rb
 - lib/bootsnap.rb
 - lib/bootsnap/bundler.rb
+- lib/bootsnap/cli.rb
+- lib/bootsnap/cli/worker_pool.rb
 - lib/bootsnap/compile_cache.rb
 - lib/bootsnap/compile_cache/iseq.rb
 - lib/bootsnap/compile_cache/yaml.rb
@@ -134,6 +138,7 @@ metadata:
   bug_tracker_uri: https://github.com/Shopify/bootsnap/issues
   changelog_uri: https://github.com/Shopify/bootsnap/blob/master/CHANGELOG.md
   source_code_uri: https://github.com/Shopify/bootsnap
+  allowed_push_host: https://rubygems.org
 post_install_message: 
 rdoc_options: []
 require_paths:
@@ -149,7 +154,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.0.2
+rubygems_version: 3.0.3
 signing_key: 
 specification_version: 4
 summary: Boot large ruby/rails apps faster