MiniRacer

Test Gem

Minimal, modern embedded V8 for Ruby.

MiniRacer provides a minimal two way bridge between the V8 JavaScript engine and Ruby.

It was created as an alternative to the excellent therubyracer, which is no longer maintained. Unlike therubyracer, mini_racer only implements a minimal bridge. This reduces the surface area making upgrading v8 much simpler and exhaustive testing simpler.

MiniRacer has an adapter for execjs so it can be used directly with Rails projects to minify assets, run babel or compile CoffeeScript.

This repository is mini_racer-csim (a fork)

This is mini_racer-csim, a private fork of mini_racer maintained for capybara-simulated. It adds browser-fidelity extensions (ES modules, realm reset, …) that capybara-simulated needs but most users do not — if you are not using capybara-simulated, use upstream mini_racer.

It stays a drop-in replacement: the gem is still loaded with require "mini_racer" and exposes the MiniRacer module, so existing code keeps working. Only the gem name (mini_racer-csim) differs.

Additions over upstream

Feature API Notes
Bytecode cache Context#compile(src, cached_data:, produce_cache:) → Script, Script#run, Script#cache_rejected? Cross-process V8 bytecode caching to skip parsing; see Bytecode cache for repeated script evaluation below
ES Module API Context#compile_module → MiniRacer::Module (#instantiate / #evaluate / #namespace / #status / #cached_data / #dispose); Context#dynamic_import_resolver= V8's ES module pipeline, import.meta.url, dynamic import()
Batched module-graph loader Context#load_module_graph(resolve:, …) Loads an ESM graph in one batched, native (C++) pass; one Module per URL shared across every load path
Realm reset Context#reset_realm Discards the user realm (globalThis) while keeping the warm isolate (browser per-navigation model); re-binds attached host functions and the host namespace
Host namespace Context.new(host_namespace: "MiniRacer") → globalThis.MiniRacer.drainMicrotasks() Opt-in JS namespace exposing an inline, rendezvous-free microtask checkpoint
GVL release on boot (automatic) Releases the Ruby GVL while the V8 thread boots the isolate

The fork is periodically rebased on upstream mini_racer to pick up V8 / libv8-node bumps and bug fixes.

Supported Ruby Versions & Troubleshooting

MiniRacer only supports non-EOL versions of Ruby. See Ruby Maintenance Branches for the list of non-EOL Rubies. If you require support for older versions of Ruby install an older version of the gem. TruffleRuby is also supported.

MiniRacer does not support

If you have a problem installing MiniRacer, please consider the following steps:

  • make sure you try the latest released version of mini_racer
  • make sure you have Rubygems >= 3.2.13 and bundler >= 2.2.13 installed: gem update --system
  • if you are using bundler
  • make sure to recompile/reinstall mini_racer and libv8-node after OS upgrades (for example via gem uninstall --all mini_racer libv8-node)
  • make sure you are on the latest patch/teeny version of a supported Ruby branch

Features

Simple eval for JavaScript

You can simply eval one or many JavaScript snippets in a shared context

context = MiniRacer::Context.new
context.eval("var adder = (a,b)=>a+b;")
puts context.eval("adder(20,22)")
# => 42

Attach global Ruby functions to your JavaScript context

You can attach one or many ruby proc that can be accessed via JavaScript

context = MiniRacer::Context.new
context.attach("math.adder", proc{|a,b| a+b})
puts context.eval("math.adder(20,22)")
# => 42

context = MiniRacer::Context.new
context.attach("array_and_hash", proc{{a: 1, b: [1, {a: 1}]}})
puts context.eval("array_and_hash()")
# => {"a" => 1, "b" => [1, {"a" => 1}]}

Return binary data from Ruby to JavaScript

Attached Ruby functions can return binary data as Uint8Array using MiniRacer::Binary:

require "digest"

context = MiniRacer::Context.new
context.attach("sha256_raw", ->(data) {
  MiniRacer::Binary.new(Digest::SHA256.digest(data))
})

# Inside JavaScript the return value is a Uint8Array
context.eval("sha256_raw('hello') instanceof Uint8Array") # => true
context.eval("sha256_raw('hello').length")                 # => 32

This is useful when you need to pass raw bytes (e.g., cryptographic digests, compressed data, binary file contents) from Ruby to JavaScript. The MiniRacer::Binary wrapper tells the bridge to serialize the data as a Uint8Array on the JavaScript side rather than a string.

GIL free JavaScript execution

The Ruby Global interpreter lock is released when scripts are executing:

context = MiniRacer::Context.new
Thread.new do
  sleep 1
  context.stop
end
context.eval("while(true){}")
# => exception is raised

This allows you to execute multiple scripts in parallel.

Timeout Support

Contexts can specify a default timeout for scripts

context = MiniRacer::Context.new(timeout: 1000)
context.eval("while(true){}")
# => exception is raised after 1 second (1000 ms)

Memory softlimit Support

Contexts can specify a memory softlimit for scripts

# terminates script if heap usage exceeds 200mb after V8 garbage collection has run
context = MiniRacer::Context.new(max_memory: 200_000_000)
context.eval("var a = new Array(10000); while(true) {a = a.concat(new Array(10000)) }")
# => V8OutOfMemoryError is raised

Rich Debugging with File Name in Stack Trace Support

You can provide filename: to #eval which will be used in stack traces produced by V8:

context = MiniRacer::Context.new
context.eval("var foo = function() {bar();}", filename: "a/foo.js")
context.eval("bar()", filename: "a/bar.js")

# JavaScript at a/bar.js:1:1: ReferenceError: bar is not defined (MiniRacer::RuntimeError)
# …

Bytecode cache for repeated script evaluation

Context#compile returns a MiniRacer::Script handle you can run multiple times, and exposes V8's bytecode cache so subsequent Contexts can skip the parse step.

In a single process — e.g. warming a Context pool from one canonical compile:

# Warm the cache once — top-level compile, opt in with produce_cache: true.
warm    = MiniRacer::Context.new
warmed  = warm.compile(File.read("bundle.js"),
                       filename:      "bundle.js",
                       produce_cache: true)
warmed.run
blob = warmed.cached_data   # ASCII-8BIT String, hold onto it in memory

# Subsequent Contexts (e.g. a per-request pool) consume the blob and skip parsing.
ctx    = MiniRacer::Context.new
script = ctx.compile(File.read("bundle.js"),
                     filename:    "bundle.js",
                     cached_data: blob)
# script.cache_rejected? is false when V8 accepted the blob.
script.run

Across processes (e.g. persisting blobs to disk), the consumer must boot from byte-identical snapshot data — two separate Snapshot.new(src) calls produce different blobs even for the same src, and V8 will then reject every cached blob. Use Snapshot#dump / Snapshot.load to share canonical bytes:

# Build the snapshot once, persist its bytes.
snap_bytes = MiniRacer::Snapshot.new(snapshot_src).dump
File.binwrite("snapshot.bin", snap_bytes)

# Every process loads the same bytes.
snap = MiniRacer::Snapshot.load(File.binread("snapshot.bin"))
ctx  = MiniRacer::Context.new(snapshot: snap)
script = ctx.compile(File.read("bundle.js"),
                     filename:    "bundle.js",
                     cached_data: File.binread("bundle.js.cache"))
script.run

produce_cache defaults to false; pass true to ask V8 for the cache blob. When the supplied cached_data is accepted, script.cached_data returns nil so callers can skip a redundant copy. When V8 produces a fresh blob (initial compile with produce_cache: true, or a rejection while produce_cache: true was also set), it returns the new bytes.

MiniRacer::V8_CACHED_DATA_VERSION_TAG exposes V8's ScriptCompiler::CachedDataVersionTag() — mix it into your cache key alongside the source hash so a libv8-node version bump invalidates stale blobs automatically. The constant is populated on first Context.new (after Platform.set_flags!), so read it after constructing at least one Context.

key = "#{Digest::SHA256.hexdigest(source)}-#{MiniRacer::V8_CACHED_DATA_VERSION_TAG}"

Notes:

  • A Script is bound to the Context that compiled it; reusing it on another Context isn't supported.
  • Script#dispose frees the underlying V8 handle eagerly. The Ruby GC finalizer does not (taking the V8 lock from a finalizer thread risks deadlock), so long-lived Contexts with many short-lived scripts accumulate handles until Context#dispose clears them.
  • produce_cache: true is only safe at the top level. From inside a host-fn callback (i.e., re-entrant compile while a JS → Ruby → JS frame is on the stack) it raises MiniRacer::RuntimeError, because V8's CreateCodeCache walks live isolate state and corrupts the parser when re-entered. Warm the cache from the top level once and pass it back via cached_data: from your callbacks.
  • Cross-process reuse is incompatible with MiniRacer::Platform.set_flags!(:single_threaded). V8's single-threaded mode embeds process-local state in the cache blob, so every cached_data is rejected when consumed in a fresh process. Same-process reuse still works under :single_threaded. If you need both cross-process reuse and :single_threaded (e.g. for fork-safety reasons), disable :single_threaded for the path that produces / consumes the cache.
  • On TruffleRuby, Script is implemented as source replay (GraalJS has no equivalent per-script bytecode cache reachable from Polyglot::InnerContext), so cached_data and produce_cache are silently ignored and cached_data always returns nil, and MiniRacer::V8_CACHED_DATA_VERSION_TAG is 0.

Fork Safety

Some Ruby web servers employ forking (for example unicorn or puma in clustered mode). V8 is not fork safe by default and sadly Ruby does not have support for fork notifications per #5446.

Since 0.6.1 mini_racer does support V8 single threaded platform mode which should remove most forking related issues. To enable run this before using MiniRacer::Context, for example in a Rails initializer:

MiniRacer::Platform.set_flags!(:single_threaded)

When using pre-fork MiniRacer::Context objects in :single_threaded mode, ensure the process only forks while MiniRacer is quiescent: no thread may be evaluating JavaScript, calling into a context, disposing/freeing a context, running a Ruby callback from JavaScript, or otherwise using MiniRacer at the instant of fork. In multi-threaded applications, guard all MiniRacer context operations and the fork itself with the same application-level lock. Forking while a MiniRacer operation is in progress can leave inherited pthread mutexes in an unusable state in the child process.

If you want to ensure your application does not leak memory after fork either:

  1. Ensure no MiniRacer::Context objects are created in the master process; or
  2. Dispose manually of all MiniRacer::Context objects prior to forking
# before fork

require "objspace"
ObjectSpace.each_object(MiniRacer::Context){|c| c.dispose}

# fork here

Threadsafe

Context usage is threadsafe

context = MiniRacer::Context.new
context.eval("counter=0; plus=()=>counter++;")

(1..10).map do
  Thread.new {
    context.eval("plus()")
  }
end.each(&:join)

puts context.eval("counter")
# => 10

Snapshots

Contexts can be created with pre-loaded snapshots:

snapshot = MiniRacer::Snapshot.new("function hello() { return 'world!'; }")

context = MiniRacer::Context.new(snapshot: snapshot)

context.eval("hello()")
# => "world!"

Snapshots can come in handy for example if you want your contexts to be pre-loaded for efficiency. It uses V8 snapshots under the hood; see this link for caveats using these, in particular:

There is an important limitation to snapshots: they can only capture V8’s heap. Any interaction from V8 with the outside is off-limits when creating the snapshot. Such interactions include:

  • defining and calling API callbacks (i.e. functions created via v8::FunctionTemplate)
  • creating typed arrays, since the backing store may be allocated outside of V8

And of course, values derived from sources such as Math.random or Date.now are fixed once the snapshot has been captured. They are no longer really random nor reflect the current time.

Also note that snapshots can be warmed up, using the warmup! method, which allows you to call functions which are otherwise lazily compiled to get them to compile right away; any side effect of your warm up code being then dismissed. More details on warming up here, and a small example:

snapshot = MiniRacer::Snapshot.new("var counter = 0; function hello() { counter++; return 'world! '; }")

snapshot.warmup!("hello()")

context = MiniRacer::Context.new(snapshot: snapshot)

context.eval("hello()")
# => "world! 1"
context.eval("counter")
# => 1

Snapshots can also be persisted to disk for faster startup:

# Save a snapshot to disk
snapshot = MiniRacer::Snapshot.new('var foo = "bar";')
File.binwrite("snapshot.bin", snapshot.dump)

# Load it back in a later process
blob = File.binread("snapshot.bin")
snapshot = MiniRacer::Snapshot.load(blob)
context = MiniRacer::Context.new(snapshot: snapshot)
context.eval("foo")
# => "bar"

Note that snapshots are architecture and V8-version specific. A snapshot created on one platform (e.g., ARM64 macOS) cannot be loaded on a different platform (e.g., x86_64 Linux). Snapshots are best used for same-machine caching or homogeneous deployment environments.

Security note: Only load snapshots from trusted sources. V8 snapshots are not designed to be safely loaded from untrusted input—malformed or malicious snapshot data may cause crashes or memory corruption.

Garbage collection

You can make the garbage collector more aggressive by defining the context with MiniRacer::Context.new(ensure_gc_after_idle: 1000). Using this will ensure V8 will run a full GC using context.low_memory_notification 1 second after the last eval on the context. Low memory notifications ensure long living contexts use minimal amounts of memory.

V8 Runtime flags

It is possible to set V8 Runtime flags:

MiniRacer::Platform.set_flags! :noconcurrent_recompilation, max_inlining_levels: 10

This can come in handy if you want to use MiniRacer with Unicorn, which doesn't seem to always appreciate V8's liberal use of threading:

MiniRacer::Platform.set_flags! :noconcurrent_recompilation, :noconcurrent_sweeping

Or else to unlock experimental features in V8, for example tail recursion optimization:

MiniRacer::Platform.set_flags! :harmony

js = <<-JS
'use strict';
var f = function f(n){
  if (n <= 0) {
    return  'foo';
  }
  return f(n - 1);
}

f(1e6);
JS

context = MiniRacer::Context.new

context.eval js
# => "foo"

The same code without the harmony runtime flag results in a MiniRacer::RuntimeError: RangeError: Maximum call stack size exceeded exception. Please refer to http://node.green/ as a reference on other harmony features.

A list of all V8 runtime flags can be found using node --v8-options, or else by perusing the V8 source code for flags (make sure to use the right version of V8).

Note that runtime flags must be set before any other operation (e.g. creating a context or a snapshot), otherwise an exception will be thrown.

Flags:

  • :expose_gc: Will expose gc() which you can run in JavaScript to issue a GC run.
  • :max_old_space_size: defaults to 1400 (megs) on 64 bit, you can restrict memory usage by limiting this.

NOTE TO READER our documentation could be awesome we could be properly documenting all the flags, they are hugely useful, if you feel like documenting a few more, PLEASE DO, PRs are welcome.

Controlling memory

When hosting v8 you may want to keep track of memory usage, use #heap_stats to get memory usage:

context = MiniRacer::Context.new
# use context
p context.heap_stats
# {:total_physical_size=>1280640,
#  :total_heap_size_executable=>4194304,
#  :total_heap_size=>3100672,
#  :used_heap_size=>1205376,
#  :heap_size_limit=>1501560832}

If you wish to dispose of a context before waiting on the GC use #dispose:

context = MiniRacer::Context.new
context.eval("let a='testing';")
context.dispose
context.eval("a = 2")
# MiniRacer::ContextDisposedError

# nothing works on the context from now on, it's a shell waiting to be disposed

A MiniRacer context can also be dumped in a heapsnapshot file using #write_heap_snapshot(file_or_io)

context = MiniRacer::Context.new
# use context
context.write_heap_snapshot("test.heapsnapshot")

This file can then be loaded in the "memory" tab of the Chrome DevTools.

Function call

This calls the function passed as first argument:

context = MiniRacer::Context.new
context.eval("function hello(name) { return `Hello, ${name}!` }")
context.call("hello", "George")
# "Hello, George!"

Performance is slightly better than running context.eval("hello('George')") since:

  • compilation of eval'd string is avoided
  • function arguments don't need to be converted to JSON

Microtask checkpoints

V8 drains its microtask queue (e.g. callbacks queued via Promise.resolve().then(...)) automatically when script execution returns to the embedder, so most code "just works":

context = MiniRacer::Context.new
context.eval(<<~JS)
  let x = 0;
  Promise.resolve().then(() => x = 99);
JS
context.eval("x")
# => 99

When JavaScript invokes a Ruby callback synchronously and you need queued microtasks to drain mid-execution — e.g. for spec-compliant ordering across a chain of synchronous dispatchEvent listeners — call context.perform_microtask_checkpoint from the callback:

context = MiniRacer::Context.new
context.attach("drain", -> { context.perform_microtask_checkpoint })
context.eval(<<~JS)
  globalThis.log = [];
  Promise.resolve().then(() => log.push("microtask"));
  log.push("before");
  drain();
  log.push("after");
JS
context.eval("log")
# => ["before", "microtask", "after"]

Without drain() the order would be ["before", "after", "microtask"] because the microtask only runs once the outermost script returns. perform_microtask_checkpoint is a thin wrapper over V8's MicrotasksScope::PerformCheckpoint.

When the drain has to happen from within JavaScript itself — for example between each listener in a synchronous dispatchEvent chain — the same checkpoint is available to JS as drainMicrotasks(). It runs inline on the V8 thread without the Ruby ↔ V8 round-trip, so no attach is required.

It is exposed through an opt-in host namespace — a single object (in the spirit of Deno's Deno or Bun's Bun) that mini_racer hangs its non-standard helpers off. Pass host_namespace: to enable it; by default nothing is injected and the global stays clean:

context = MiniRacer::Context.new(host_namespace: "MiniRacer")
context.eval(<<~JS)
  globalThis.log = [];
  Promise.resolve().then(() => log.push("microtask"));
  log.push("before");
  MiniRacer.drainMicrotasks();
  log.push("after");
JS
context.eval("log")
# => ["before", "microtask", "after"]

host_namespace: accepts a String (the global name to use — it must be a valid JavaScript identifier), true (the default name "MiniRacer"), or nil/false (the default — inject nothing). The namespace object is defined non-enumerable so it does not appear in Object.keys(globalThis), while its methods are ordinary properties discoverable via Object.keys(MiniRacer). Like perform_microtask_checkpoint, drainMicrotasks() is a no-op while a microtask checkpoint is already in progress, and it lets watchdog/out-of-memory termination propagate to the enclosing eval/call. (The host namespace is V8-only; it is not installed on the TruffleRuby backend.)

ES modules

Context#compile_module exposes V8's ES module API for code that uses import / export syntax. Unlike eval (which only accepts script-level syntax), modules can have static imports that resolve to other modules and expose named exports through a real Module Namespace Object.

context = MiniRacer::Context.new

dep  = context.compile_module("export const base = 10", filename: "dep.js")
main = context.compile_module(<<~JS, filename: "main.js")
  import { base } from 'dep'
  export const doubled = base * 2
JS

main.instantiate {|specifier, referrer| dep }  # called once per static import
dep.evaluate
main.evaluate

main.namespace  # => {"doubled" => 20}
  • Context#compile_module(source, filename:) — parses the source as a module; the returned MiniRacer::Module is bound to its Context. The filename is also exposed to the module as import.meta.url.
  • Module#instantiate { |specifier, referrer_url| ... } — walks the static import graph. The resolver block is called once per import declaration with the raw specifier string and the importing module's filename, so relative specifiers (./foo, ../bar) can be resolved against the referrer. It must return another MiniRacer::Module (typically from a per-Context cache). Imports can also be resolved lazily from inside the block via further Context#compile_module calls.
  • Module#evaluate — runs the module body. Returns the evaluation result (nil for the typical export const … shape). Modules with top-level await raise MiniRacer::RuntimeError for now.
  • Module#namespace — returns the Module Namespace Object as a Hash ({ "default" => …, "namedExport" => … }). Available after instantiate succeeds; evaluate populates the values.
  • Module#status — one of :uninstantiated, :instantiating, :instantiated, :evaluating, :evaluated, :errored.
  • Module#dispose / Module#disposed? — eager handle release, mirroring the convention used elsewhere.
  • Context#dynamic_import_resolver = proc { |specifier, referrer_url| ... } — handler for JS import(...) expressions. The proc must return a MiniRacer::Module (already instantiated; evaluate is driven for you if pending). Set to nil to reject all dynamic imports. Drain the microtask queue with Context#perform_microtask_checkpoint to see the result in a .then callback or after await.
context.dynamic_import_resolver = ->(spec, _ref) { cache.fetch(spec) }
context.eval(%(import('dep').then(ns => globalThis.r = ns.x)), filename: 'caller.js')
context.perform_microtask_checkpoint
context.eval('globalThis.r')  # => 42

Notes:

  • A Module is bound to the Context that compiled it; resolvers must return modules from the same Context.
  • Module#dispose frees the underlying V8 handle eagerly. The Ruby GC finalizer does not (taking the V8 lock from a finalizer thread risks deadlock), so long-lived Contexts with many short-lived modules accumulate handles until Context#dispose clears them.
  • Top-level await is not yet supported; evaluate raises if the module's evaluation promise stays pending after the microtask drain.
  • On TruffleRuby, Context#compile_module raises NotImplementedError — GraalJS has its own module-loading mechanism that doesn't map onto this handle-based API. PRs to bridge are welcome.

Performance

The bench folder contains benchmark.

Benchmark minification of Discourse application.js (both minified and non-minified)

MiniRacer outperforms node when minifying assets via execjs.

  • MiniRacer version 0.1.9
  • node version 6.10
  • therubyracer version 0.12.2
$ bundle exec ruby bench.rb mini_racer
Benching with mini_racer
mini_racer minify discourse_app.js 9292.72063ms
mini_racer minify discourse_app_minified.js 11799.850171ms
mini_racer minify discourse_app.js twice (2 threads) 10269.570797ms

sam@ubuntu exec_js_uglify % bundle exec ruby bench.rb node
Benching with node
node minify discourse_app.js 13302.715484ms
node minify discourse_app_minified.js 18100.761243ms
node minify discourse_app.js twice (2 threads) 14383.600207000001ms

sam@ubuntu exec_js_uglify % bundle exec ruby bench.rb therubyracer
Benching with therubyracer
therubyracer minify discourse_app.js 171683.01867700001ms
therubyracer minify discourse_app_minified.js 143138.88492ms
therubyracer minify discourse_app.js twice (2 threads) NEVER FINISH

Killed: 9

The huge performance disparity (MiniRacer is 10x faster) is due to MiniRacer running latest version of V8. In July 2016 there is a queued upgrade to therubyracer which should bring some of the perf inline.

Note how the global interpreter lock release leads to 2 threads doing the same work taking the same wall time as 1 thread.

As a rule MiniRacer strives to always support and depend on the latest stable version of libv8.

Source Maps

MiniRacer can fully support source maps but must be configured correctly to do so. Check out this example for a working implementation.

Installation

Add this line to your application's Gemfile:

gem "mini_racer"

And then execute:

$ bundle

Or install it yourself as:

```terminal
$ gem install mini_racer

Note using v8.h and compiling MiniRacer requires a C++20 capable compiler. gcc >= 12.2 and Xcode >= 13 are, at the time of writing, known to work.

Similar Projects

therubyracer

  • https://github.com/cowboyd/therubyracer
  • Most comprehensive bridge available
  • Provides the ability to "eval" JavaScript
  • Provides the ability to invoke Ruby code from JavaScript
  • Hold references to JavaScript objects and methods in your Ruby code
  • Hold references to Ruby objects and methods in JavaScript code
  • Uses libv8, so installation is fast
  • Supports timeouts for JavaScript execution
  • Does not release global interpreter lock, so performance is constrained to a single thread
  • Currently (May 2016) only supports v8 version 3.16.14 (Released approx November 2013), plans to upgrade by July 2016
  • Supports execjs

v8eval

  • https://github.com/sony/v8eval
  • Provides the ability to "eval" JavaScript using the latest V8 engine
  • Does not depend on the libv8 gem, installation can take 10-20 mins as V8 needs to be downloaded and compiled.
  • Does not release global interpreter lock when executing JavaScript
  • Does not allow you to invoke Ruby code from JavaScript
  • Multi runtime support due to SWIG based bindings
  • Supports a JavaScript debugger
  • Does not support timeouts for JavaScript execution
  • No support for execjs (can not be used with Rails uglifier and coffeescript gems)

therubyrhino

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/rubyjs/mini_racer. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the Contributor Covenant code of conduct.

License

The gem is available as open source under the terms of the MIT License.