Ruby 2.6.0 Released

Posted by naruse on 25 Dec 2018

We are pleased to announce the release of Ruby 2.6.0.

It introduces a number of new features and performance improvements, most notably:

• A new JIT compiler.
• The RubyVM::AbstractSyntaxTree module.

JIT [Experimental]

Ruby 2.6 introduces an initial implementation of a JIT (Just-In-Time) compiler.

The JIT compiler aims to improve the performance of Ruby programs. Unlike traditional JIT compilers which operate in-process, Ruby’s JIT compiler writes out C code to disk and spawns a common C compiler to generate native code. For more details about it, see the MJIT organization by Vladimir Makarov.

In order to enable the JIT compiler, specify --jit on the command line or in the $RUBYOPT environment variable. Specifying --jit-verbose=1 will cause the JIT compiler to print additional information. Read the output of ruby --help or the documentation for other options.

The JIT compiler is supported when Ruby is built by GCC, Clang, or Microsoft VC++, which needs to be available at runtime.

As of Ruby 2.6.0, we have achieved 1.7x faster performance compared to Ruby 2.5 on a CPU-intensive, non-trivial benchmark called Optcarrot. However, it is still experimental and many other memory-intensive workloads like Rails applications might not benefit from it at the moment. For more details, see Ruby 2.6 JIT - Progress and Future.

Stay tuned for the new age of Ruby’s performance.

RubyVM::AbstractSyntaxTree [Experimental]

Ruby 2.6 introduces the RubyVM::AbstractSyntaxTree module. Future compatibility of this module is not guaranteed.

This module has a parse method, which parses the given string as Ruby code and returns the AST (Abstract Syntax Tree) nodes of the code. The parse_file method opens and parses the given file as Ruby code and returns AST nodes.

The RubyVM::AbstractSyntaxTree::Node class is also introduced. You can get source location and children nodes from Node objects. This feature is experimental.

Other Notable New Features

• Add an alias of Kernel#yield_self named #then. [Feature #14594]

• Constant names may start with a non-ASCII capital letter. [Feature #13770]

• Introduce endless ranges. [Feature #12912]

  An endless range, (1..), works as if it has no end. Here are some typical use cases:

  ary[1..]                          # identical to ary[1..-1] without magical -1
  (1..).each {|index| ... }         # enumerates values starting from index 1
  ary.zip(1..) {|elem, index| ... } # ary.each.with_index(1) { ... }
  

• Add Enumerable#chain and Enumerator#+. [Feature #15144]

• Add function composition operators << and >> to Proc and Method. [Feature #6284]

  f = proc{|x| x + 2}
  g = proc{|x| x * 3}
  (f << g).call(3) # -> 11; identical to f(g(3))
  (f >> g).call(3) # -> 15; identical to g(f(3))
  

• Add Binding#source_location. [Feature #14230]

  This method returns the source location of the binding, a 2-element array of __FILE__ and __LINE__. Technically speaking, this is identical to eval("[__FILE__, __LINE__]", binding). However, we are planning to change this behavior so that Kernel#eval ignores binding’s source location [Bug #4352]. As such, it is recommended to use Binding#source_location instead of Kernel#eval.

• Add an exception: option to Kernel#system which causes it to raise an exception on failure instead of returning false. [Feature #14386]

• Add a oneshot mode to Coverage. [Feature#15022]

  • This mode checks “whether each line was executed at least once or not”, instead of “how many times each line was executed”. A hook for each line is fired only once, and once it is fired the hook flag will be removed, i.e., it runs with zero overhead.
  • Add oneshot_lines: keyword argument to Coverage.start.
  • Add stop: and clear: keyword arguments to Coverage.result. If clear is true, it clears the counters to zero. If stop is true, it disables coverage measurement.
  • Coverage.line_stub is a simple helper function that creates the “stub” of line coverage from a given source code.

• Add FileUtils#cp_lr. It works just like cp_r but links instead of copies. [Feature #4189]

Performance improvements

• Speed up Proc#call by removing the temporary allocation for $SAFE. [Feature #14318]

  We have observed a 1.4x performance improvement in the lc_fizzbuzz benchmark that calls Proc#call numerous times. [Bug #10212]

• Speed up block.call when block is passed in as a block parameter. [Feature #14330]

  Combined with improvements around block handling introduced in Ruby 2.5, block evaluation now performs 2.6x faster in a micro-benchmark in Ruby 2.6. [Feature #14045]

• Transient Heap (theap) is introduced. [Bug #14858] [Feature #14989]

  theap is a managed heap for short-living memory objects which are pointed to by specific classes (Array, Hash, Object, and Struct). Making small and short-living Hash objects is 2x faster. With rdoc benchmark, we observed 6-7% performance improvement.

• Native implementations (arm32, arm64, ppc64le, win32, win64, x86, amd64) of coroutines to improve context switching performance of Fiber significantly. [Feature #14739]

  Fiber.yield and Fiber#resume is about 5x faster on 64-bit Linux. Fiber intensive programs can expect up to 5% improvement overall.

Other notable changes since 2.5

• $SAFE is now a process global state and it can be set back to 0. [Feature #14250]

• Passing safe_level to ERB.new is deprecated. trim_mode and eoutvar arguments have been changed to keyword arguments. [Feature #14256]

• Unicode support is updated to version 11. We have plans to add support for Unicode version 12 and 12.1 in a future TEENY release of Ruby 2.6. This will include support for the new Japanese era.

• Merge RubyGems 3.0.1. The --ri and --rdoc options have been removed. Please use the --document and --no-document options instead.

• Bundler is now installed as a default gem.

• In exception handling blocks, else without rescue now causes a syntax error. [EXPERIMENTAL][Feature #14606]

See NEWS or the commit logs for more details.

With those changes, 6437 files changed, 231471 insertions(+), 98498 deletions(-) since Ruby 2.5.0!

Merry Christmas, Happy Holidays, and enjoy programming with Ruby 2.6!

Known Problem

(This section was added at January 28, 2019.)

• Net::Protocol::BufferedIO#write raises NoMethodError when sending large multi-byte string

Download

• https://cache.ruby-lang.org/pub/ruby/2.6/ruby-2.6.0.tar.gz

  SIZE:   16687800 bytes
  SHA1:   c95f4e86e21390270dad3ebb94491fd42ee2ce69
  SHA256: f3c35b924a11c88ff111f0956ded3cdc12c90c04b72b266ac61076d3697fc072
  SHA512: 01f886b0c0782a06315c4a46414e9f2b66ee634ba4349c8e0697f511793ae3c56d2ad3cad6563f2b0fdcedf0ff3eba51b9afab907e7e1ac243475772f8688382
  

• https://cache.ruby-lang.org/pub/ruby/2.6/ruby-2.6.0.zip

  SIZE:   20582054 bytes
  SHA1:   a804e63d18da12107e1d101918a3d8f4c5462a27
  SHA256: 8a4fb6ca58202495c9682cb88effd804398bd0ef023e3e36f001ca88d8b5855a
  SHA512: 16d66ec4a2c6a2e928d5b50e094a5efa481ac6e4d5ed77459d351ef19fe692aa59b68307e3e25229eec5f30ae2f9adae2663bafe9c9d44bfb45d3833d77839d4
  

• https://cache.ruby-lang.org/pub/ruby/2.6/ruby-2.6.0.tar.bz2

  SIZE:   14585856 bytes
  SHA1:   b8638eb806efbf7b6af87b24ccc6ad915f262318
  SHA256: c89ca663ad9a6238f4b1ec4d04c7dff630560c6e6eca6d30857c4d394f01a599
  SHA512: ca3daf9acf11d3db2900af21b66231bd1f025427a9d2212b35f6137ca03f77f57171ddfdb99022c8c8bcd730ff92a7a4af54e8a2a770a67d8e16c5807aa391f1
  

• https://cache.ruby-lang.org/pub/ruby/2.6/ruby-2.6.0.tar.xz

  SIZE:   11918536 bytes
  SHA1:   9ddaeba3505d2855460c8c653159fc0ac8928c0f
  SHA256: acb00f04374899ba8ee74bbbcb9b35c5c6b1fd229f1876554ee76f0f1710ff5f
  SHA512: c56eaf85ef7b79deb34ee4590b143c07f4fc83eb79775290761aee5a7c63374659613538a41f25706ed6e19e49d5c67a1014c24d17f29948294c7abd0b0fcea8