NEWS for Ruby 4.1.0

This document is a list of user-visible feature changes since the 4.0.0 release, except for bug fixes.

Note that each entry is kept to a minimum, see links for details.

Language changes

super in a module method that was called via super from a refinement method now works correctly. [[Feature #22309]]
Module#clone and Module#dup no longer rewrite the lexical scope of copied methods. Constants and class variables resolve through the original class, consistent with inheritance and mixins. [[Feature #21981]]
alias and Module#alias_method in a module emit a deprecation warning when the original method is not a method of the module or its ancestors and is only found through the fallback to Object. The fallback will be removed in Ruby 4.3, and such an alias will raise a NameError. [[Bug #22276]]
alias and Module#alias_method emit a deprecation warning when the original method is defined in a prepended module. This behavior will be removed in Ruby 4.3, and alias lookup will start at the origin class (after prepended modules). [[Bug #22273]]

Core classes updates

Note: We're only listing outstanding class updates.

C API

* `RB_NOGVL_PENDING_INTR_FAIL` is added as a flag for `rb_nogvl`.
`rb_nogvl` does not enter the blocking region (and does not call the
given function), returning `0`, if the current thread has pending
interrupts, including interrupts masked by `Thread.handle_interrupt`.
As with the existing skip path, `errno` is `0` because the function was
never called.
* `RHASH_TBL` is deprecated.  [[Feature #22232]]

Array

* `Array#pack` accepts new formats `R` and `r` for unsigned and signed
LEB128 encoded integers. [[Feature #21785]]
* `Array#pack` accepts new formats `x!` and `@!` to align the current
offset to a byte boundary or to the ABI alignment of another
directive. [[Feature #22185]]

Dir

* `Dir.glob` supports `***/` for recursive matching that follows
directory symbolic links, stopping at links to ancestor directories.
[[Feature #17548]]

ENV

* `ENV.fetch_values` is added. It returns an array of values for the
given names, raising `KeyError` for missing names unless a block is
given. [[Feature #21781]]

Fiber::Scheduler

* The `fiber_interrupt` hook is now required. Schedulers which do not
implement it can no longer be installed with `Fiber.set_scheduler`.

* `GC.stat` accepts a `process:` keyword. `GC.stat` and
`GC.stat(process: false)` report the current Ractor's object space.
With the default collector, `GC.stat(process: true)` reports cumulative
collection counts and measured marking/sweeping CPU time across Ractors,
including history from destroyed object spaces, without requesting a
stop-the-world snapshot. It also reports the `:page_pool_arenas`,
`:page_pool_arenas_freed`, `:page_pool_total_pages` and
`:page_pool_discarded_pages` keys, which the Ractor data does not
include. `GC.count` and `GC.total_time` retain their current-Ractor
behavior.
* `GC.stat` reports a new key, `:global_gc_count`: the number of
stop-the-world global garbage collection cycles that collected the
current Ractor's heap, whichever Ractor started them. The new key
`:global_gc_invocation_count` reports the global collections that the
current Ractor started. `GC.stat(:global_gc_count, process: true)`
reports the number of global collections in the process, including
those started by terminated Ractors.
* A global garbage collection is not counted in `:major_gc_count`.
`:count` equals `:minor_gc_count` plus `:major_gc_count` plus
`:global_gc_count` in both the Ractor and the process data. In the
process data each global collection counts once. When another Ractor's
object space exists, including one left by a terminated Ractor that is
not yet absorbed, or one was absorbed since the last global collection,
`GC.start` runs a global collection by default, so it increments
`:global_gc_count` rather than `:major_gc_count`; pass `global: false`
for a local major collection.
* Global compaction no longer counts overlapping CPU intervals in scalar GC
timing statistics. Profiler wall-time intervals are unchanged.

Hash

* `Hash.ruby2_keywords_hash?` and `Hash.ruby2_keywords_hash` are
deprecated and will be removed in Ruby 4.5. [[Feature #22205]]

Integer

* `Integer#bit_count` is added. It returns the number of `1` bits in the
binary representation of a non-negative integer (its population count).
[[Feature #20163]]

IO::Buffer

* `read`, `write`, `pread`, and `pwrite` now perform one IO operation using
`(offset, length)`, where `length` is the maximum transfer size. Short
transfers are returned directly.

* The corresponding fiber scheduler hooks are no longer experimental and
now use the same `(offset, length)` argument order and single-operation
semantics.

Kernel

* `Kernel#autoload_relative` and `Module#autoload_relative` are added.
These methods work like `autoload`, but resolve the file path relative
to the file where the method is called, similar to `require_relative`.
This makes it easier to autoload constants from files in the same
directory without hardcoding absolute paths or manipulating `$LOAD_PATH`.
[[Feature #15330]]

MatchData

* `MatchData#integer_at` is added.  It converts the matched substring to
integer and return the result.  [[Feature #21932]]

Module

* `Module#descendants` is added.  It returns an array of classes and
modules that have the receiver in their ancestors.  [[Feature #9779]]
* `Module#ruby2_keywords` and top-level `ruby2_keywords` are
deprecated and will be removed in Ruby 4.4.  [[Feature #22205]]
* `Module#method_defined?` now accepts a third optional argument to also
match private methods.  [[Feature #22297]]

ObjectSpace

* `ObjectSpace._id2ref` was removed.  [[Feature #22135]]

Proc

* `Proc#refined` is added. It returns a new `Proc` that behaves like the
receiver but with the refinements activated by the given modules
in effect inside its body, without affecting the original `Proc`.
[[Feature #22097]]
* `Proc#ruby2_keywords` is deprecated and will be removed in Ruby 4.4.
[[Feature #22205]]

Ractor

* `Ractor::Port#receive`, `Ractor.receive` and `Ractor.select` accept a
`timeout:` keyword argument (in seconds) and return `nil` once it passes.
`timeout: 0` never blocks: it takes a message if one is already queued
and returns `nil` otherwise.  [[Feature #22255]]

* `Ractor#alive?` is added. It returns `false` once the ractor has
terminated. [[Feature #22441]]

Range

* `Range#clamp` is added. It returns a new `Range` instance whose
begin and end values are clamped to the given bounds. [[Feature #22175]]

Regexp

* All instances of `Regexp` are now frozen, not just literals.
Subclasses of `Regexp` are not frozen for compatibility.
[[Feature #8948]]

Set

* A deprecated behavior, `Set#to_set`, `Range#to_set`, and
`Enumerable#to_set` accepting arguments, was removed.  [[Feature #21390]]

String

* `String#unpack` and `String#unpack1` accept new formats `R` and `r`
for unsigned and signed LEB128 encoded integers. [[Feature #21785]]
* `String#unpack` and `String#unpack1` accept a new format `^` that
returns the current offset.  Useful when combined with variable
width formats like LEB128. [[Feature #21796]]
* `String#unpack` and `String#unpack1` accept new formats `x!` and
`@!` to align the current offset to a byte boundary or to the ABI
alignment of another directive. [[Feature #22185]]
* Basic bit operations are added. `String#bit_get`, `String#bit_set?`,
`String#bit_set`, `String#bit_clear`, `String#bit_flip` and
`String#bit_count` handle individual bits, and `String#bitwise_not`,
`String#bitwise_and`, `String#bitwise_or`, `String#bitwise_xor`
(with their `!` variants) handle whole strings. [[Feature #22118]]
* `String#tr` now accept a Hash for multi-character replacement.
[[Feature #22238]]

Symbol

* `Symbol#to_s` now returns a frozen string. [[Feature #22137]]

Thread

* Assigning `nil` with `Thread#thread_variable_set` now deletes the
thread-local variable. [[Feature #19333]]

Thread::Backtrace::Location

* `Thread::Backtrace::Location#source_range` is added. It returns a
`Ruby::SourceRange` for the Ruby expression associated with the frame.

Stdlib updates

Psych

* An experimental libfyaml backend has been added.  It is only
enabled when psych is built with `--enable-libfyaml`, and the
default libyaml backend remains unchanged otherwise.  It is not
supported on Windows.  [[GH-psych #805]]

Ruby::Prism

* `Ruby::Prism` is added.  It is the prism parser built into the
interpreter, i.e., the very parser that parsed the running program,
independent of the prism gem.  It is autoloaded.
`Method#syntax_tree` and others return its nodes.  Its API may change
between Ruby minor versions; use the `prism` gem unless you need the
parser of the running interpreter.  [[Bug #22379]]

The following bundled gems are added.

We only list stdlib changes that are notable feature changes.

Other changes are listed in the following sections. We also listed release history from the previous bundled version that is Ruby 3.4.0 if it has GitHub releases.

The following bundled gems are promoted from default gems.

prism 1.9.0
- 1.7.0 to [v1.8.0][prism-v1.8.0], [v1.8.1][prism-v1.8.1], [v1.9.0][prism-v1.9.0]
tsort 0.2.0
win32-registry 0.1.2

The following bundled gems are removed.

net-ftp 0.3.9
net-pop 0.1.2

They are still available on rubygems.org and can be installed with gem install. [[Feature #21835]]

The following default gem is added.

The following default gems are updated.

RubyGems 4.1.0.beta2
- 4.0.3 to [v4.0.4][RubyGems-v4.0.4], [v4.0.5][RubyGems-v4.0.5], [v4.0.6][RubyGems-v4.0.6], [v4.0.7][RubyGems-v4.0.7], [v4.0.8][RubyGems-v4.0.8], [v4.0.9][RubyGems-v4.0.9], [v4.0.10][RubyGems-v4.0.10], [v4.0.11][RubyGems-v4.0.11], [v4.0.12][RubyGems-v4.0.12], [v4.0.13][RubyGems-v4.0.13], [v4.0.14][RubyGems-v4.0.14], [v4.0.15][RubyGems-v4.0.15], [v4.0.16][RubyGems-v4.0.16], [v4.0.17][RubyGems-v4.0.17], [v4.0.18][RubyGems-v4.0.18], [v4.0.19][RubyGems-v4.0.19], [v4.0.20][RubyGems-v4.0.20], [v4.0.21][RubyGems-v4.0.21], [v4.0.22][RubyGems-v4.0.22], [v4.0.23][RubyGems-v4.0.23], [v4.1.0.beta1][RubyGems-v4.1.0.beta1], [v4.1.0.beta2][RubyGems-v4.1.0.beta2]
bundler 4.1.0.beta2
- 4.0.3 to [v4.0.4][bundler-v4.0.4], [v4.0.5][bundler-v4.0.5], [v4.0.6][bundler-v4.0.6], [v4.0.7][bundler-v4.0.7], [v4.0.8][bundler-v4.0.8], [v4.0.9][bundler-v4.0.9], [v4.0.10][bundler-v4.0.10], [v4.0.11][bundler-v4.0.11], [v4.0.12][bundler-v4.0.12], [v4.0.13][bundler-v4.0.13], [v4.0.14][bundler-v4.0.14], [v4.0.15][bundler-v4.0.15], [v4.0.16][bundler-v4.0.16], [v4.0.17][bundler-v4.0.17]
date 3.6.0
- 3.5.1 to [v3.6.0][date-v3.6.0]
delegate 0.7.0
- 0.6.1 to [v0.7.0][delegate-v0.7.0]
digest 3.3.0
- 3.2.1 to [v3.3.0][digest-v3.3.0]
erb 6.1.0
- 6.0.1 to [v6.0.1.1][erb-v6.0.1.1], [v6.0.2][erb-v6.0.2], [v6.0.3][erb-v6.0.3], [v6.0.4][erb-v6.0.4], [v6.0.5][erb-v6.0.5], [v6.0.6][erb-v6.0.6], [v6.0.7][erb-v6.0.7]
error_highlight 0.7.2
etc 1.5.0
- 1.4.6 to [v1.5.0][etc-v1.5.0]
fileutils 1.8.1
- 1.8.0 to [v1.8.1][fileutils-v1.8.1]
io-console 0.9.4
- 0.8.2 to [v0.9.0][io-console-v0.9.0], [v0.9.1][io-console-v0.9.1], [v0.9.2][io-console-v0.9.2], [v0.9.3][io-console-v0.9.3], [v0.9.4][io-console-v0.9.4]
io-wait 999.999.999
ipaddr 1.3.0
- 1.2.8 to [v1.2.9][ipaddr-v1.2.9], [v1.3.0][ipaddr-v1.3.0]
json 3.1.0
- 2.18.0 to [v2.18.1][json-v2.18.1], [v2.19.0][json-v2.19.0], [v2.19.1][json-v2.19.1], [v2.19.2][json-v2.19.2], [v2.19.3][json-v2.19.3], [v2.19.4][json-v2.19.4], [v2.19.5][json-v2.19.5], [v2.19.6][json-v2.19.6], [v2.19.7][json-v2.19.7], [v2.19.8][json-v2.19.8], [v2.19.9][json-v2.19.9], [v2.20.0][json-v2.20.0], [v2.21.0][json-v2.21.0], [v2.21.2][json-v2.21.2], [v3.0.0.rc1][json-v3.0.0.rc1], [v3.0.0][json-v3.0.0]
net-http 0.10.0
- 0.9.1 to [v0.10.0][net-http-v0.10.0]
net-protocol 0.4.0
- 0.2.2 to [v0.3.0][net-protocol-v0.3.0], [v0.4.0][net-protocol-v0.4.0]
open3 0.3.0
- 0.2.1 to [v0.3.0][open3-v0.3.0]
openssl 4.0.2
- 4.0.0 to [v4.0.1][openssl-v4.0.1], [v4.0.2][openssl-v4.0.2]
pp 0.6.4
- 0.6.3 to [v0.6.4][pp-v0.6.4]
prettyprint 0.2.1
- 0.2.0 to [v0.2.1][prettyprint-v0.2.1]
psych 5.5.0
- 5.3.1 to [v5.4.0][psych-v5.4.0], [v5.5.0][psych-v5.5.0]
resolv 0.8.1
- 0.7.0 to [v0.7.1][resolv-v0.7.1], [v0.7.2][resolv-v0.7.2], [v0.8.0][resolv-v0.8.0], [v0.8.1][resolv-v0.8.1]
stringio 3.2.1.dev
strscan 3.1.9.dev
- 3.1.6 to [v3.1.7][strscan-v3.1.7], [v3.1.8][strscan-v3.1.8]
syntax_suggest 3.0.0
tempfile 0.3.2
- 0.3.1 to [v0.3.2][tempfile-v0.3.2]
time 0.5.0
- 0.4.2 to [v0.5.0][time-v0.5.0]
timeout 0.6.1
- 0.6.0 to [v0.6.1][timeout-v0.6.1]
tmpdir 0.3.2
- 0.3.1 to [v0.3.2][tmpdir-v0.3.2]
uri 1.1.2
- 1.1.1 to [v1.1.2][uri-v1.1.2]
zlib 3.2.4
- 3.2.2 to [v3.2.3][zlib-v3.2.3], [v3.2.4][zlib-v3.2.4]

The following bundled gems are updated.

minitest 6.0.6
power_assert 3.1.0
- 3.0.1 to [v3.1.0][power_assert-v3.1.0]
rake 13.4.2
- 13.3.1 to [v13.4.0][rake-v13.4.0], [v13.4.1][rake-v13.4.1], [v13.4.2][rake-v13.4.2]
test-unit 3.7.8
- 3.7.5 to [3.7.6][test-unit-3.7.6], [3.7.7][test-unit-3.7.7], [3.7.8][test-unit-3.7.8]
rss 0.3.3
- 0.3.2 to [0.3.3][rss-0.3.3]
net-imap 0.6.8
- 0.6.2 to [v0.6.3][net-imap-v0.6.3], [v0.6.4][net-imap-v0.6.4], [v0.6.4.1][net-imap-v0.6.4.1], [v0.6.5][net-imap-v0.6.5], [v0.6.6][net-imap-v0.6.6], [v0.6.7][net-imap-v0.6.7], [v0.6.8][net-imap-v0.6.8]
net-smtp 0.5.2
- 0.5.1 to [v0.5.2][net-smtp-v0.5.2]
rbs 4.2.0
- 3.10.0 to [v3.10.1][rbs-v3.10.1], [v3.10.2][rbs-v3.10.2], [v3.10.3][rbs-v3.10.3], [v3.10.4][rbs-v3.10.4], [v4.0.0.dev.1][rbs-v4.0.0.dev.1], [v4.0.0.dev.2][rbs-v4.0.0.dev.2], [v4.0.0.dev.3][rbs-v4.0.0.dev.3], [v4.0.0.dev.4][rbs-v4.0.0.dev.4], [v4.0.0.dev.5][rbs-v4.0.0.dev.5], [v4.0.0][rbs-v4.0.0], [v4.0.1.dev.1][rbs-v4.0.1.dev.1], [v4.0.1.dev.2][rbs-v4.0.1.dev.2], [v4.0.1][rbs-v4.0.1], [v4.0.2][rbs-v4.0.2], [v4.0.3][rbs-v4.0.3], [v4.1.0.pre.1][rbs-v4.1.0.pre.1], [v4.1.0.pre.2][rbs-v4.1.0.pre.2], [v4.1.0][rbs-v4.1.0], [v4.1.1.pre.1][rbs-v4.1.1.pre.1], [v4.1.1][rbs-v4.1.1], [v4.1.2][rbs-v4.1.2], [v4.1.3][rbs-v4.1.3], [v4.2.0.pre.1][rbs-v4.2.0.pre.1], [v4.2.0][rbs-v4.2.0]
typeprof 0.33.2
mutex_m 0.3.0
bigdecimal 4.1.3
- 4.0.1 to [v4.1.0][bigdecimal-v4.1.0], [v4.1.1][bigdecimal-v4.1.1], [v4.1.2][bigdecimal-v4.1.2], [v4.1.3][bigdecimal-v4.1.3]
resolv-replace 0.2.0
- 0.1.1 to [v0.2.0][resolv-replace-v0.2.0]
nkf 0.3.0
- 0.2.0 to [v0.3.0][nkf-v0.3.0]
syslog 0.4.0
- 0.3.0 to [v0.4.0][syslog-v0.4.0]
csv 3.3.6
- 3.3.5 to [v3.3.6][csv-v3.3.6]
repl_type_completor 0.1.16
- 0.1.12 to [v0.1.13][repl_type_completor-v0.1.13], [v0.1.14][repl_type_completor-v0.1.14], [v0.1.15][repl_type_completor-v0.1.15], [v0.1.16][repl_type_completor-v0.1.16]
pstore 0.2.1
- 0.2.0 to [v0.2.1][pstore-v0.2.1]
rdoc 8.1.0
- 7.0.3 to [v7.0.4][rdoc-v7.0.4], [v7.1.0][rdoc-v7.1.0], [v7.2.0][rdoc-v7.2.0], [v8.0.0][rdoc-v8.0.0], [v8.1.0][rdoc-v8.1.0]
win32ole 1.9.3
- 1.9.2 to [v1.9.3][win32ole-v1.9.3]
irb 1.18.0
- 1.16.0 to [v1.17.0][irb-v1.17.0], [v1.18.0][irb-v1.18.0]
reline 0.7.0
- 0.6.3 to [v0.7.0][reline-v0.7.0]

RubyGems and Bundler

Ruby 4.0 bundled RubyGems and Bundler version 4. see the following links for details.

Supported platforms

Support code for the following platforms has been removed. Their vendors or distributors have ended support, there is no environment to test Ruby on them, and none of them has a platform maintainer.

* AmigaOS
* AtheOS
* BSD/OS (BSDi)
* ESIX
* EWS-UX (NEC EWS4800)
* FreeBSD, NetBSD and OpenBSD with a.out binaries
* GNU/kFreeBSD
* GNU/kOpenSolaris
* HI-UX/MPP
* HP-UX on the HP 9000 Series 300
* Interix (Windows Services for UNIX)
* IRIX
* MirOS BSD
* NeXTSTEP, OpenStep and Rhapsody
* OSF/1 (Tru64 UNIX)
* QNX Neutrino
* SunOS 4 (Solaris, i.e. SunOS 5, is unaffected)
* System V Release 4
* UNICOS/mp (Cray X1)
* UX/PDS

Support code for the following old versions has been removed.

* Cygwin 1.3 and earlier
* Solaris 9 and earlier
* Linux 2.6 and earlier (RHEL 5)

Windows 10 1703 or later no longer needs the LongPathsEnabled registry value to use paths longer than 260 characters. This applies to any process running the interpreter, including a program which embeds libruby. Each path component is still limited to 255 characters, and a child process still starts with the MAX_PATH limited current directory. [[Bug #18947]]
On 64-bit Windows with MSVC (mswin), String, Array and MatchData can now be 2GiB or longer. Their lengths used to be long, which is 32 bits there. [[Feature #22400]]
Building Ruby with MSVC now requires Visual Studio 2017 version 15.8 (_MSC_VER 1915) or later.
Ruby built with MSVC now requires Windows 10 version 1809 (build 17763) or Windows Server 2019 or later. Earlier Windows 10 releases and Windows Server 2016 are no longer supported.
Ruby on macOS now requires Mac OS X 10.6 or later. Mac OS X 10.5 and PowerPC Macs are no longer supported.

Compatibility issues

A class or module can now be modified only by the Ractor which created it, its owner. Defining, removing or undefining methods, alias, changing visibility, include/prepend, Module#refine, defining or removing constants, registering an autoload, writing the class's own instance variables and class variables, Module#freeze and Module#set_temporary_name raise Ractor::IsolationError in any other Ractor. Reading is unchanged. Everything defined at boot or by the main Ractor, required libraries included, is owned by the main Ractor, so a non-main Ractor can no longer monkey-patch it; and since defining a constant in a foreign class is prohibited, it can not define a top-level class or module name either. In exchange a Ractor has full use of the classes it creates itself, including unshareable constant, instance variable and class variable values, which not even the main Ractor could do before. See doc/language/ractor.md. [[Feature #22226]]
Kernel#at_exit and END {} now raise Ractor::IsolationError when called in a non-main Ractor. Previously the registered handler ran in the main Ractor at process exit, which was confusing. [[Feature #22139]]
Ractor.make_shareable now raises Ractor::Error for an IO object, and Ractor.shareable? returns false for it. A frozen IO used to be shareable, but almost all of its methods raise FrozenError and it can still refer to unshareable objects through the members of rb_io_t.
Random.urandom now raises RangeError for every size too large for a String, as Random#bytes does. A size that still fit in an unsigned long used to raise ArgumentError. [[Feature #22400]]
Struct#[] and Struct#[]= now raise IndexError for every out-of-range Integer index within the range of Array indices. An index past the int range used to raise RangeError. [[Feature #22400]]

Stdlib compatibility issues

prism is now a bundled gem. Under Bundler, require "prism" needs prism in the Gemfile. Use Ruby::Prism for the parser of the running interpreter. [[Bug #22379]]
Etc.getlogin on Windows now returns the login name determined when the process starts, from the USER or USERNAME environment variable or GetUserName(). It used to follow later changes to ENV['USER'].

Socket

* On Windows, a connection that is refused or unreachable now raises the
matching `Errno` class as soon as Winsock reports it, instead of
`Errno::ETIMEDOUT` once the whole `connect_timeout` has passed. Code
rescuing `Errno::ETIMEDOUT` there has to rescue the real error instead.

* On Windows, `BasicSocket#getsockopt(:SOCKET, :ERROR)` now reports an
errno as it does on the other platforms, instead of the raw WinSock
error code. Code comparing it with a `WSAE*` value has to compare it
with the matching `Errno::*::Errno` instead.

[[Bug #18661]]

C API updates

Embedded TypedData

The RUBY_TYPED_EMBEDDABLE flag is now public and documented and can be used by C extensions. It allows allocating C structs directly into Ruby object slots, which reduces pointer chasing, and in some case memory usage. See the C extension documentation for details. [[Feature #21853]]
Added new C23 inspired allocator functions, that takes the previous memory size. This allow the Ruby GC to better keep track of memory usage, improving its heuristics. It also improves the performance of system allocators that support C23 free_sized.

However, it is important to note that passing an incorrect size to these function is undefined behavior and may result in crashes or memory leaks.
- ruby_xfree_sized(void *ptr, size_t size)
- ruby_xrealloc_sized(void *ptr, size_t newsiz, size_t oldsiz)
- ruby_xrealloc2_sized(void *ptr, size_t newelems, size_t newsiz, size_t oldelems)
[[Feature #21861]]

Ractor-scoped GC address registration

rb_gc_register_address and rb_global_variable now register the address with the calling Ractor, and only that Ractor's GC marks the stored object.

Any Ractor may register an address, but the value stored through it must be a special constant, a shareable object, or an unshareable object owned by the registering Ractor. Storing another Ractor's unshareable object can result in the object being freed while the address still refers to it (use-after-free).

If the address has process lifetime (a static VALUE), register it from the main Ractor or keep the stored values shareable.

When the registering Ractor is joined with Ractor#value, remaining registrations move to the joining Ractor; otherwise they move to the main Ractor once the dead Ractor is collected.

[[Feature #22277]]

Type for lengths and indices

rb_len_t and rb_ulen_t are added as the types of the lengths and indices of String, Array and MatchData, with LEN2NUM, NUM2LEN, ULEN2NUM, rb_len2int, PRIdLEN and its siblings, RB_LEN_MAX, RB_LEN_MIN, RB_ULEN_MAX, SIZEOF_RB_LEN_T and HAVE_RB_LEN_T. RbConfig::SIZEOF["rb_len_t"] gives its size to an extconf.rb. They are long, except on 64-bit mswin, where they are 64 bits. The C API that carries such lengths, such as RSTRING_LEN and rb_range_beg_len, uses them. [[Feature #22400]]
On 64-bit mswin this changes the ABI, so extensions have to be rebuilt. An extension that passes a long * to an output parameter such as rb_range_beg_len no longer compiles there, and one that keeps a length in long truncates it at 2GiB. Keep such lengths in rb_len_t, or convert them with rb_len2long. An extension that also supports older versions can check HAVE_RB_LEN_T and fall back to typedef long rb_len_t. [[Feature #22400]]
On mswin, extensions are now built with incompatible pointer types (warning C4133) as errors. Besides a long * passed where an rb_len_t * is expected, this rejects any other mismatched pointer that used to compile with a warning. [[Feature #22400]]
rb_len2long is added to convert an rb_len_t to long, raising RangeError via the new rb_out_of_long when it does not fit, for extensions that keep a length in long. [[Feature #22400]]

Atomic operations

RUBY_ATOMIC_SIZE_LOAD is added as the size_t counterpart of RUBY_ATOMIC_LOAD, alongside the existing RUBY_ATOMIC_SIZE_ADD, RUBY_ATOMIC_SIZE_SUB, RUBY_ATOMIC_SIZE_CAS and RUBY_ATOMIC_SIZE_EXCHANGE.

Removed APIs

The following APIs, which have been deprecated for many years, are removed. [[Feature #21768]]

old postponed job functions,
untyped data object type/functions,
old APIs to allocate a data object,
taintedness/trustedness enums/macros,
rb_gc_force_recycle function,
rb_iterate function,
and some functions and constants for internal use.

Implementation improvements

The default garbage collector has been switched from a freelist allocator to a bump pointer allocator. [[PR #17201]]
The error_highlight, did_you_mean, and syntax_suggest gems are now loaded lazily on the first error display instead of at interpreter boot, which reduces startup time. Process.warmup loads them eagerly so that pre-forking servers keep their code in copy-on-write shared memory. [[Feature #21951]]

Ractor

A lot of work has gone into making Ractors more stable, performant, and usable. These improvements bring Ractor implementation closer to leaving experimental status.

The default GC now runs per Ractor: each Ractor collects its own heap on its own thread without stopping the others, and a stop-the-world collection only runs when it is really needed (explicit full GC.start, shareable-object growth, reclaiming dead Ractors' heaps). Allocation-heavy Ractor programs now scale like forked processes.
GC.start(global: false) can be used when multiple Ractors are running to force a Ractor-local GC. By default, GC.start runs a global GC (all Ractors) like before. This can also be triggered with GC.start(global: true). The global keyword argument has no effect when a single Ractor is running.

Visible behavior changes:
Ractor#value returns the value only once; a second call raises Ractor::Error.
GC.disable/GC.enable act as per-Ractor holds on a process-wide switch: one Ractor's GC.enable no longer overrides another Ractor's GC.disable.
ObjectSpace.each_object enumerates the calling Ractor's own objects plus other Ractors' shareable objects (ObjectSpace.dump_all still covers everything).
ObjectSpace.define_finalizer on another Ractor's object raises Ractor::IsolationError.
Ractor#monitor now sends an Array naming the Ractor and what happened to it, [ractor, :exited] or [ractor, :aborted], where it used to send the bare Symbol :exited or :aborted. Several Ractors can then report to one port and the receiver still knows which one finished. The Array is built for the receiving Ractor, so watching many Ractors leaves no shareable objects behind.
```
r = Ractor.new { :ok }
r.monitor(port = Ractor::Port.new)
port.receive #=> [r, :exited]
```
One port can therefore watch a whole group, which is all a supervisor needs:
```
workers.each { |r| r.monitor port }

until workers.empty?
  r, status = port.receive
  workers.delete(r)
  workers << restart(r) if status == :aborted
end
```

M:N thread scheduler

The scheduler scales with the number of waiters and of Ractors, where it used to walk a list or take one lock for all of them:
A timed wait sits in a hierarchical timer wheel rather than on a list sorted by deadline, which was inserted into by a linear scan.
An fd stays armed in the backend between waits, instead of being added before each wait and removed after each wake.
The io-wait bookkeeping is sharded by fd, rather than serialized on one lock across every fd.
A timed wait on an fd rides the scheduler instead of going to a blocking region, which cost a native thread handoff per wait.
A context switch, and leaving or rejoining the shared pool, no longer take the scheduler's global lock.
The RUBY_MN_THREADS environment variable now runs from no M:N scheduling at all to all of it. -1 is new: a Ractor's threads have been M:N since the scheduler was added, with no way to turn that off. 0 and 1 are unchanged.

main thread the main Ractor's other threads a Ractor's threads

-1 1:1 1:1 1:1

0 or unset 1:1 1:1 M:N

1 1:1 M:N M:N

2 M:N M:N M:N
RUBY_MN_THREADS=2 is new. The main thread is resumed like any other M:N thread rather than woken on a native thread of its own, which costs an order of magnitude more. It pays off when the main thread drives the work, and does nothing for one that only starts other threads and waits.

The main thread is then no longer bound to one OS thread, which is what the M:N scheduler already meant for every other thread:
A C extension that keeps state per OS thread has to call rb_thread_lock_native_thread().
What must run on the process's initial thread does not work at all, pinning included: macOS AppKit and CFRunLoop, and hosts that embed Ruby and return into their own main loop.
The OS thread name is no longer set from the Ruby thread for M:N threads: one native thread runs many of them over its life. Thread#name= was already skipped for the same reason.

	main thread	the main Ractor's other threads	a Ractor's threads
`-1`	1:1	1:1	1:1
`0` or unset	1:1	1:1	M:N
`1`	1:1	M:N	M:N
`2`	M:N	M:N	M:N

JIT

ZJIT

Status
- ZJIT is now faster than YJIT on many benchmarks, but we expect that ZJIT can be still slower than YJIT on larger benchmarks or applications as of Ruby 4.1.0 preview releases. Stay tuned for the Ruby 4.1.0 release.
Major optimizations
- Method inlining: small Ruby methods, including methods that yield to a block, are inlined into their callers, so that the optimizer can see across method boundaries.
- Lightweight frames: method calls in JIT code no longer write a full VM control frame. Frame metadata used by backtraces and GC is materialized only when needed.
- Optimization pass: a new linear scan register allocator, load-store optimization, more constant folding, and redundant guard elimination.
- GC fastpath: allocation of objects is inlined into JIT code, skipping a call into the GC on the fast path.
More Ruby code is compiled
- Calls with blocks or &block arguments, super, splats, keyword arguments, ... forwarding, and post-required parameters.
- Specialized code for many core methods, e.g. Float arithmetic, Array#[], Array#length, Array#empty?, Integer#[], String#byteslice, String#to_sym, Class#allocate, and new on Struct classes.
- Instance variable reads and shareable constants are now compiled while multiple Ractors are running.
Command-line options
- --zjit-stats=FILE writes stats as JSON when FILE ends with .json.
Ruby API
- RubyVM::ZJIT.exit_locations and RubyVM::ZJIT.dump_exit_locations are removed along with the Stackprof output of --zjit-trace-exits.

YJIT

YJIT now supports Windows (mswin) on x64 CPUs. Configure with win32\configure.bat --enable-yjit, which requires rustc with the x86_64-pc-windows-msvc target. Only the release build is supported, and MinGW is not. [[Feature #18439]]
Add RubyVM::YJIT.max_compile_time_ns and RubyVM::YJIT.max_compile_time_ns= to give YJIT a compilation time budget. When RubyVM::YJIT.total_compile_time_ns exceeds it, YJIT pauses compilation until the budget is raised, which spreads the warmup overhead over a longer period. [[Feature #22236]]