- ruby-core - ml.ruby-lang.org

[ruby-core:124833] [Ruby Bug#21882] IO::Buffer#locked leaves the buffer locked when the block raises
by hanazuki (Kasumi Hanazuki) 09 Jul '26

09 Jul '26

Issue #21882 has been reported by hanazuki (Kasumi Hanazuki). ---------------------------------------- Bug #21882: IO::Buffer#locked leaves the buffer locked when the block raises https://bugs.ruby-lang.org/issues/21882 * Author: hanazuki (Kasumi Hanazuki) * Status: Open * ruby -v: ruby 4.0.1 (2026-01-13 revision e04267a14b) +PRISM [x86_64-linux] * Backport: 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN, 4.0: UNKNOWN ---------------------------------------- `IO::Buffer#locked` should unlock the buffer even after the block raises an exception. ```ruby buf = IO::Buffer.new(100) buf.locked { fail } rescue nil buf.locked { p :ok } # expected to print :ok, but actually raises a LockedError ``` -- https://bugs.ruby-lang.org/

5 6

[ruby-core:125821] [Ruby Bug#22127] parse.y regexp crash on invalid encoding
by kddnewton (Kevin Newton) 09 Jul '26

09 Jul '26

Issue #22127 has been reported by kddnewton (Kevin Newton). ---------------------------------------- Bug #22127: parse.y regexp crash on invalid encoding https://bugs.ruby-lang.org/issues/22127 * Author: kddnewton (Kevin Newton) * Status: Open * Backport: 3.3: UNKNOWN, 3.4: UNKNOWN, 4.0: UNKNOWN ---------------------------------------- ``` ./ruby --parser=parse.y -e '/#{"\xcd"}/' ``` results in a segfault -- https://bugs.ruby-lang.org/

3 2

[ruby-core:125931] [Ruby Bug#22179] Coverage: redefined methods silently disappear from method Coverage.result after GC
by sferik (Erik Berlin) 09 Jul '26

09 Jul '26

Issue #22179 has been reported by sferik (Erik Berlin). ---------------------------------------- Bug #22179: Coverage: redefined methods silently disappear from method Coverage.result after GC https://bugs.ruby-lang.org/issues/22179 * Author: sferik (Erik Berlin) * Status: Open * Backport: 3.3: UNKNOWN, 3.4: UNKNOWN, 4.0: UNKNOWN ---------------------------------------- When method coverage is enabled, a method that has been redefined vanishes from `Coverage.result` once GC runs. It is not reported as uncovered (count 0), it is missing altogether, so the set of methods reported for a file changes nondeterministically depending on GC timing. ##### Reproduction target.rb: ```ruby def foo 1 end def foo 2 end foo ``` run.rb: ```ruby require "coverage" Coverage.start(methods: true) require_relative "./target" GC.start if ENV["FORCE_GC"] pp Coverage.result.find { |f, _| f.end_with?("target.rb") }.last[:methods] ``` Without GC, both definitions are reported: ``` $ ruby run.rb {[Object, :foo, 5, 0, 7, 3] => 1, [Object, :foo, 1, 0, 3, 3] => 0} ``` With a GC before `Coverage.result`, the first definition is gone: ``` $ FORCE_GC=1 ruby run.rb {[Object, :foo, 5, 0, 7, 3] => 1} ``` The explicit `GC.start` is only for determinism. In real programs any GC that happens to run between the redefinition and `Coverage.result` has the same effect, which is how this was noticed in the wild. I’ve looked into why this is happening and I believe it's because method coverage is not stored per file the way line and branch coverage are, it's reconstructed at result time by walking the heap for method entries (`method_coverage_i` via `rb_objspace_each_objects` in ext/coverage/coverage.c). A shadowed definition that was never called has no entry in `me2counter` and exists only as a garbage method entry after redefinition, so once GC collects it, the heap walk cannot see it and the definition is silently forgotten. This issue was discovered when it was reported as [a bug in SimpleCov](https://github.com/simplecov-ruby/simplecov/issues/1217), of which I am a maintainer. Reproduced on 3.2.11, 3.3.11, 3.4.10, and 4.0.5. -- https://bugs.ruby-lang.org/

2 2

[ruby-core:125986] [Ruby Misc#22180] Provide official Windows (mswin) binary packages and a version manager
by naruse (Yui NARUSE) 09 Jul '26

09 Jul '26

Issue #22180 has been updated by naruse (Yui NARUSE). > It looks like packaging vcruntime140.dll is dubious... vcruntime is a part of libc on Windows and Visual C++. Though libc on Unix usually consists from a single binary, Windows separates it into Universal CRT and VCRuntime. ucrt provides main part of libc, and VCRuntime provides C compiler related things for example exception hendling and so on. Just remove vcruntime140.dll and assume the user installed VC++ Redistributable package is rather reasonable. ---------------------------------------- Misc #22180: Provide official Windows (mswin) binary packages and a version manager https://bugs.ruby-lang.org/issues/22180#change-117963 * Author: hsbt (Hiroshi SHIBATA) * Status: Assigned * Assignee: hsbt (Hiroshi SHIBATA) ---------------------------------------- Japanese version of this proposal: https://gist.github.com/hsbt/552b01f2aac616af6f8056148407900a ## Background Ruby used to ship official Windows binaries. https://ftp.ruby-lang.org/pub/ruby/binaries/mswin32/ still has mswin32 zips from the 1.6 to 1.9-preview era, and they were discontinued around 1.9. One reason is that a bare zip was not usable as a distribution. https://bugs.ruby-lang.org/issues/999 reports that the `ruby-1.9.1-preview1` mswin32 zip was missing the OpenSSL and zlib DLLs, so `gem update` did not work at all. Since then, installing Ruby on Windows has depended on third-party distributions. The mswin platform, which is what `ruby/ruby` CI actually builds and tests with MSVC, has had no end-user distribution at all. https://bugs.ruby-lang.org/issues/19325 shows how the current state is perceived by Windows users. The technical ground has changed. Over the last few months I fixed all the tests that remained broken on mswin myself: `nmake check`, `test-bundler` and `test-bundled-gems` are green now, and `nmake exam`, which runs all of the test suites, completes as well. The build manages its C dependencies with a pinned vcpkg manifest (`vcpkg.json`), and CI covers VS 2022, VS 2026 and `windows-11-arm`. MSVC ABI (`vcruntime140`) has been stable since 2015. The missing piece is distribution, not build quality. Timing also matters. In the era of writing code with AI, the people who need a working Ruby on Windows are increasingly ordinary users who let a coding assistant set up the environment for them, and consumer AI tools operate natively on Windows, not through WSL. An assistant working unattended needs a runtime that installs without elevation and without GUI dialogs, exposes a machine-readable version index, and is relocatable so that installs are deterministic. Unattended provisioning is already the norm in CI (`setup-ruby` exists precisely for this), and AI assistants bring the same requirement to end-user machines. Python reached the same conclusion with PEP 773, described below. Ruby has no official answer to this on Windows today. ## Proposal I propose to provide an official Windows distribution based on the mswin platform, made of three components. All of them, including the test-suite fixes above, are work I implemented and verified myself. Nothing below is a design on paper. **1. `ruby/ruby`: a `binary-package` target.** A new nmake-only target packages the installed tree together with the vcpkg runtime DLLs, the VC runtime and the license terms of everything redistributed into a relocatable zip, with build-machine specific paths scrubbed from `rbconfig.rb`. Thanks to `LOAD_RELATIVE` the tree runs from any location. I will submit the pull request shortly. This is implemented and verified. The package passes a smoke test with `PATH` restricted to `System32` (openssl, fiddle, psych, zlib and RubyGems all work), and C extensions compile with VS Build Tools. In a survey of the 500 most-downloaded gems on rubygems.org forced to build from source, 21 of the 44 gems with a native extension built and loaded. All 23 failures were gem-side or environment-side issues (gcc-only code, POSIX-only source, missing third-party libraries), none caused by the package or by ruby itself, with identical results on 3.3.11, 3.4.10 and 4.0.5. **2. `ruby/actions`: building and publishing.** Automation builds the zips and publishes them together with sha256 checksums: daily builds of master and release builds of the maintained series. Released series (3.3, 3.4, 4.0) are built from their release tarballs with the same packaging script applied externally, so no change to the stable branches is required. This also enables re-packaging revisions (`-1`, `-2`) when a bundled dependency such as OpenSSL needs a security update without a new Ruby release. This part is already working too: in addition to master, zips of 3.3.11, 3.4.10 and 4.0.5 have been built and verified this way, and they are the packages the gem survey above was run on. **3. `rbmanager`: a version manager for Windows.** The same shape as Python's install manager (PEP 773, https://peps.python.org/pep-0773/) A single native executable (the `rb` command) installs a zip from a URL or a local path, extracts it under `%LOCALAPPDATA%`, switches versions with an NTFS junction, and needs one user-level `PATH` entry and no elevation. A working prototype written in C# is already complete and verified on a real machine: ``` > rb install ruby-4.1.0-x64-mswin64_140.zip Extracting ruby-4.1.0-x64-mswin64_140 ... Installed ruby-4.1.0-x64-mswin64_140 > ruby -v ruby 4.1.0dev (2026-06-29T22:22:44Z master 2ce6b8e54a) +PRISM [x64-mswin64_140] > rb list ruby-3.3.11-x64-mswin64_140 ruby-3.4.10-x64-mswin64_140 ruby-4.0.5-x64-mswin64_140 * ruby-4.1.0-x64-mswin64_140 > rb use 4.0 Now using ruby-4.0.5-x64-mswin64_140 > ruby -v ruby 4.0.5 (2026-05-20 revision 64336ffd0e) +PRISM [x64-mswin64_140] ``` The example installs from a local zip today. Once the official version index described below is published, `rb install 4.1` will resolve the version, download the zip and verify its checksum by itself. ## What building on the platform vendor's toolchain enables This proposal is complementary to the existing gcc-based distributions and does not compete with them. Some capabilities are decided by third-party vendors rather than by Ruby, because their SDKs assume the MSVC toolchain and ABI, and those become reachable with an official MSVC-built runtime. Machine-learning runtime assets become directly reusable: the prebuilt Windows packages of ONNX Runtime, libtorch and OpenCV are MSVC binaries with C++ ABIs, and CUDA's `nvcc` accepts only `cl.exe` as its host compiler on Windows, so Ruby extensions binding GPU and ML runtimes need an MSVC-built Ruby to link cleanly. PDB symbols connect Ruby to the standard Windows diagnostics stack: WinDbg and Visual Studio can debug a mixed stack of Ruby and vendor DLLs, WER minidumps from end users become analyzable, and publishing official symbol packages becomes possible for the first time. And extension authors can pull C dependencies from the same vcpkg ecosystem that `ruby/ruby` itself already uses through `vcpkg.json`. ## What I ask to decide First, make these official distribution artifacts: announce them on the www.ruby-lang.org download page and host the zips, checksums and a static version index (the equivalent of pymanager's `index-windows.json`) on ruby-lang.org infrastructure. GitHub Releases can serve as the initial hosting during the preview period. Second, sign the published binaries with a project identity (for example SignPath Foundation issues OSS code-signing certificates) rather than an individual certificate. Third, host `rbmanager` as a repository under the `ruby` organization. The concrete zip layout and naming rule are implementation details. I will work them out with @nobu and document the result in `doc/distribution/windows.md`, so they are not something to decide on this ticket. ## Scope and non-goals This proposal does not replace or deprecate any existing third-party distribution. It adds an official channel for the platform the core team itself builds and tests. I already operate the official snap distribution and the entire release pipeline including security releases, and these artifacts ride on the same automation and the same operator, so this proposal does not change the current maintenance level. Pure-Ruby gems work with no toolchain. Building a C extension requires VS Build Tools, which Microsoft has licensed for building open-source dependencies without a paid Visual Studio license since 2022, and a `ridk`-like bootstrap (`win32/install-buildtools.cmd` is the seed) is follow-up work. Precompiled `x64-mswin64` gems across the ecosystem are long-term work and not a prerequisite. The survey above identified 12 gems that already support Windows with `x64-mingw-ucrt` precompiled binaries and only lack a working MSVC source path, a concrete and small upstream fix list. That list includes nokogiri and ffi, so applications depending on them (Rails, for example) do not work on this distribution yet. The fixes are tractable: I have already verified a one-line fix for msgpack locally and will send these upstream. TLS trust anchors are not bundled in the zip. `rbmanager` provisions certificates from the Windows certificate store at install time, and I will separately propose `OpenSSL::X509::Store#add_store` support for the OpenSSL 3.2+ `org.openssl.winstore://` store in ruby/openssl as the permanent solution. x64 comes first. arm64 (`arm64-mswin64`) already builds in CI and will follow once its test suite is green. ## Roadmap The rollout is incremental rather than tied to a single date. Zips for the released series (3.3, 3.4, 4.0) will be published progressively for their existing releases. For 4.1, nightly zips of master are published continuously, and a zip follows each preview release as soon as it is out. Only the GA of `rbmanager` and the distribution mechanism as a whole is aligned with the Ruby 4.1 release in December. -- https://bugs.ruby-lang.org/

1 0

[ruby-core:123795] [Ruby Bug#21685] Unnecessary context-switching, especially bad on multi-core machines.
by jpl-coconut (Jacob Lacouture) 09 Jul '26

09 Jul '26

Issue #21685 has been reported by jpl-coconut (Jacob Lacouture). ---------------------------------------- Bug #21685: Unnecessary context-switching, especially bad on multi-core machines. https://bugs.ruby-lang.org/issues/21685 * Author: jpl-coconut (Jacob Lacouture) * Status: Open * ruby -v: ruby 3.4.7 (2025-10-08 revision 7a5688e2a2) +PRISM [aarch64-linux] * Backport: 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN ---------------------------------------- While debugging a performance issue in a large rails application, I wrote a minimal microbenchmark that reproduces the issue. [[here]](https://gist.github.com/jpl-coconut/cb3679ce885eb578e1071c4b3a525d5c) I was surprised to see that the benchmark takes ~3.6sec on a single-core machine, and ~36sec **(10x slower) on a machine with 2 or more cores** . Initially I thought this was a bug in the implementation of Thread::Queue, but soon realized it relates to how the ruby reschedules threads around system calls. I prepared a fix in [[this branch]](https://github.com/jpl-coconut/ruby/tree/deferred_thread_wait) which is based off ruby 3.4.7. I can apply the fix to a different branch or to master if that's helpful. The fix simply defers suspending the thread until the syscall has been running for some short interval. I chose 100usec initially, but this could easily be made configurable. I pasted raw benchmark results below from a single run (though I did many runs and the results are stable). My CPU is an Apple M4. After the fix: - Single-core performance improves by 55%, from 3.6sec to 2sec. - Adding cores causes performance to be flat (at 2sec), rather than getting 10x slower. - Multi-core context-switch count reduces by 99.995%, from 1.4 million to ~80 - system_time/user_time ratio drops from (1.2 - 1.6) to 0.65 Here are the benchmark results before my change: ``` # time taskset --cpu-list 1 ./ruby qtest_simple.rb voluntary_ctxt_switches: 1140773 nonvoluntary_ctxt_switches: 9487 real 0m3.619s user 0m1.653s sys 0m1.950s # time taskset --cpu-list 1,2 ./ruby qtest_simple.rb voluntary_ctxt_switches: 1400110 nonvoluntary_ctxt_switches: 3 real 0m36.223s user 0m9.380s sys 0m14.927s ``` And after: ``` # time taskset --cpu-list 1 ./ruby qtest_simple.rb voluntary_ctxt_switches: 88 nonvoluntary_ctxt_switches: 899 real 0m2.031s user 0m1.209s sys 0m0.743s # time taskset --cpu-list 1,2 ./ruby qtest_simple.rb voluntary_ctxt_switches: 75 nonvoluntary_ctxt_switches: 8 real 0m2.062s user 0m1.279s sys 0m0.783s ``` I was concerned these results might still be reflective of a bug in Thread::Queue, so I also came up with a repro that doesn't rely on it. That one is [[here]](https://gist.github.com/jpl-coconut/aa14e59354abf98f808daaf39baa9a72). Results summary: - Single-core performance improves (this time by only 30%) - Multi-core penalty drops from 4x to 0. - No change to context-switching rates. - system_time/user_time ratio drops from (0.5-1) to 0.15 Before fix: ``` # time taskset --cpu-list 1 ./ruby mbenchmark.rb voluntary_ctxt_switches: 60 real 0m0.336s user 0m0.211s sys 0m0.118s # time taskset --cpu-list 1,2 ./ruby mbenchmark.rb voluntary_ctxt_switches: 60 real 0m1.424s user 0m0.468s sys 0m0.496s ``` After fix: ``` # time taskset --cpu-list 1 ./ruby mbenchmark.rb voluntary_ctxt_switches: 59 real 0m0.241s user 0m0.202s sys 0m0.032s # time taskset --cpu-list 1,2 ./ruby mbenchmark.rb voluntary_ctxt_switches: 60 real 0m0.238s user 0m0.195s sys 0m0.035s ``` -- https://bugs.ruby-lang.org/

7 18

[ruby-core:125714] [Ruby Misc#22107] DevMeeting-2026-07-09
by mame (Yusuke Endoh) 08 Jul '26

08 Jul '26

Issue #22107 has been reported by mame (Yusuke Endoh). ---------------------------------------- Misc #22107: DevMeeting-2026-07-09 https://bugs.ruby-lang.org/issues/22107 * Author: mame (Yusuke Endoh) * Status: Open ---------------------------------------- # The next dev meeting **Date: 2026/07/09 13:00-17:00** (JST) Log: *TBD* - Dev meeting *IS NOT* a decision-making place. All decisions should be done at the bug tracker. - Dev meeting is a place we can ask Matz, nobu, nurse and other developers directly. - Matz is a very busy person. Take this opportunity to ask him. If you can not attend, other attendees can ask instead of you (if attendees can understand your issue). - We will write a record of the discussion in the file or to each ticket in English. - All activities are best-effort (keep in mind that most of us are volunteer developers). - The date, time and place of the meeting are scheduled according to when/where we can reserve Matz's time. - *DO NOT* discuss then on this ticket, please. # Call for agenda items If you have a ticket that you want matz and committers to discuss, please post it into this ticket in the following format: ``` * [Ticket ref] Ticket title (your name) * Comment (A summary of the ticket, why you put this ticket here, what point should be discussed, etc.) ``` Example: ``` * [Feature #14609] `Kernel#p` without args shows the receiver (ko1) * I feel this feature is very useful and some people say :+1: so let discuss this feature. ``` - It is recommended to add a comment by 2026/07/06. We hold a preparatory meeting to create an agenda a few days before the dev-meeting. - The format is strict. We'll use [this script to automatically create an markdown-style agenda](https://gist.github.com/mame/b0390509ce1491b43610b9ebb665eb86). We may ignore a comment that does not follow the format. - Your comment is mandatory. We cannot read all discussion of the ticket in a limited time. We appreciate it if you could write a short summary and update from a previous discussion. -- https://bugs.ruby-lang.org/

12 15

[ruby-core:125684] [Ruby Feature#22102] C API to obtain a C string from a Ruby String
by rhenium (Kazuki Yamaguchi) 07 Jul '26

07 Jul '26

Issue #22102 has been reported by rhenium (Kazuki Yamaguchi). ---------------------------------------- Feature #22102: C API to obtain a C string from a Ruby String https://bugs.ruby-lang.org/issues/22102 * Author: rhenium (Kazuki Yamaguchi) * Status: Open ---------------------------------------- Extracted from https://bugs.ruby-lang.org/issues/19315#note-35 It would be nice to have a utility function whose sole purpose is to obtain a C string (NUL-terminated and containing no NUL bytes) from a Ruby String. I suggest adding something like the following to the public C API: ```c /** * Returns a pointer to the string contents as a C string. This checks that the * string does not contain embedded NUL bytes and that it is properly * NUL-terminated. * * @param[in] str String in question. * @exception rb_eArgError String contents include a NUL byte in the middle. * @return Pointer to its contents. * @pre `str` must be an instance of ::RString. */ const char *rb_str_cstr(VALUE str); ``` While extensions can currently use `StringValueCStr(val)`, there are cases where it's inconvenient. - The argument `val` needs to be addressable and modifiable, even if a type conversion is not needed. ```c const char *ptr = StringValueCStr(rb_ary_entry(ary, 0)); // Won't compile ``` - Repeating `StringValue*()` calls on the same object to perform different checks is awkward. ```c // Let's say, `val_a` and `val_b` are passed as an input, and the types are unknown StringValueCStr(val_a); // The result is unusable because the next line can invalidate it const char *b = StringValueCStr(val_b); // `val_b.to_str` can mutate `val_a` const char *a = StringValueCStr(val_a); // So you have to check `val_a` again ``` -- https://bugs.ruby-lang.org/

4 6

[ruby-core:125963] [Ruby Feature#16757] Add intersection to Range
by Eregon (Benoit Daloze) 07 Jul '26

07 Jul '26

Issue #16757 has been updated by Eregon (Benoit Daloze). I think we should add this, see discussion in #22175. ---------------------------------------- Feature #16757: Add intersection to Range https://bugs.ruby-lang.org/issues/16757#change-117939 * Author: stuyam (Stuart Yamartino) * Status: Open ---------------------------------------- It would be great if there was a way to do an intersection with ranges. I wrote a method recently to solve this but it gets complicated and doesn't solve for all edge cases. The example I was using it for was getting the intersection of two date ranges. I was using it to calculate pricing discounts for overlapping dates for a reservation system. I would propose something like: ``` ruby (Date.new(2020, 04, 03)..Date.new(2020, 04, 20)) & (Date.new(2020, 04, 15)..Date.new(2020, 04, 30)) => Wed, 15 Apr 2020..Mon, 20 Apr 2020 ``` There are a handful of array methods that you can just convert a range to and do on an arrays rather than the ranges themselves, like so: ``` ruby (Date.new(2020, 04, 03)..Date.new(2020, 04, 20)).to_a & (Date.new(2020, 04, 15)..Date.new(2020, 04, 30)).to_a => [Wed, 15 Apr 2020, Thu, 16 Apr 2020, Fri, 17 Apr 2020, Sat, 18 Apr 2020, Sun, 19 Apr 2020, Mon, 20 Apr 2020] ``` There are a few issues with this however: 1. Performance: creating the ranges into arrays and doing array calculations can me slower based on the span of the range. 2. Returning a range: if you want to go back to a range, you need to convert the resulting array back to a range, which involves min and maxing the array and handling nil edge cases. 3. Infinite ranges: Now that there are endless and beginless ranges added in ruby 2.6 and 2.7 respectively, you can not convert an infinite range to an array and therefore can't do an intersection without more complex logic. I think the added infinite range support makes it a good reason to add range intersections since that is a feature you can't accomplish with just an array. It looks like there was an [Array#intersection alias](https://bugs.ruby-lang.org/issues/16155) added for Array#& recently. I would propose Range#& and Range#intersection to have parity with Array and Set methods. -- https://bugs.ruby-lang.org/

1 0

[ruby-core:125958] [Ruby Misc#22180] Provide official Windows (mswin) binary packages and a version manager
by alanwu (Alan Wu) 07 Jul '26

07 Jul '26

Issue #22180 has been updated by alanwu (Alan Wu). It looks like packaging `vcruntime140.dll` is dubious, because it gets security updates all the time, even if it's not a direct violation of Microsoft's redistribution terms: https://www.quora.com/Am-I-legally-allowed-in-C-Visual-Studio-to-distribute… Curious if you've considered linking with [UCRT](https://learn.microsoft.com/en-us/cpp/porting/upgrade-your-code-to-th… instead? If it works, that's one fewer DLL to ship, and one fewer thing to install for users on Windows 10 and up. ---------------------------------------- Misc #22180: Provide official Windows (mswin) binary packages and a version manager https://bugs.ruby-lang.org/issues/22180#change-117932 * Author: hsbt (Hiroshi SHIBATA) * Status: Assigned * Assignee: hsbt (Hiroshi SHIBATA) ---------------------------------------- Japanese version of this proposal: https://gist.github.com/hsbt/552b01f2aac616af6f8056148407900a ## Background Ruby used to ship official Windows binaries. https://ftp.ruby-lang.org/pub/ruby/binaries/mswin32/ still has mswin32 zips from the 1.6 to 1.9-preview era, and they were discontinued around 1.9. One reason is that a bare zip was not usable as a distribution. https://bugs.ruby-lang.org/issues/999 reports that the `ruby-1.9.1-preview1` mswin32 zip was missing the OpenSSL and zlib DLLs, so `gem update` did not work at all. Since then, installing Ruby on Windows has depended on third-party distributions. The mswin platform, which is what `ruby/ruby` CI actually builds and tests with MSVC, has had no end-user distribution at all. https://bugs.ruby-lang.org/issues/19325 shows how the current state is perceived by Windows users. The technical ground has changed. Over the last few months I fixed all the tests that remained broken on mswin myself: `nmake check`, `test-bundler` and `test-bundled-gems` are green now, and `nmake exam`, which runs all of the test suites, completes as well. The build manages its C dependencies with a pinned vcpkg manifest (`vcpkg.json`), and CI covers VS 2022, VS 2026 and `windows-11-arm`. MSVC ABI (`vcruntime140`) has been stable since 2015. The missing piece is distribution, not build quality. Timing also matters. In the era of writing code with AI, the people who need a working Ruby on Windows are increasingly ordinary users who let a coding assistant set up the environment for them, and consumer AI tools operate natively on Windows, not through WSL. An assistant working unattended needs a runtime that installs without elevation and without GUI dialogs, exposes a machine-readable version index, and is relocatable so that installs are deterministic. Unattended provisioning is already the norm in CI (`setup-ruby` exists precisely for this), and AI assistants bring the same requirement to end-user machines. Python reached the same conclusion with PEP 773, described below. Ruby has no official answer to this on Windows today. ## Proposal I propose to provide an official Windows distribution based on the mswin platform, made of three components. All of them, including the test-suite fixes above, are work I implemented and verified myself. Nothing below is a design on paper. **1. `ruby/ruby`: a `binary-package` target.** A new nmake-only target packages the installed tree together with the vcpkg runtime DLLs, the VC runtime and the license terms of everything redistributed into a relocatable zip, with build-machine specific paths scrubbed from `rbconfig.rb`. Thanks to `LOAD_RELATIVE` the tree runs from any location. I will submit the pull request shortly. This is implemented and verified. The package passes a smoke test with `PATH` restricted to `System32` (openssl, fiddle, psych, zlib and RubyGems all work), and C extensions compile with VS Build Tools. In a survey of the 500 most-downloaded gems on rubygems.org forced to build from source, 21 of the 44 gems with a native extension built and loaded. All 23 failures were gem-side or environment-side issues (gcc-only code, POSIX-only source, missing third-party libraries), none caused by the package or by ruby itself, with identical results on 3.3.11, 3.4.10 and 4.0.5. **2. `ruby/actions`: building and publishing.** Automation builds the zips and publishes them together with sha256 checksums: daily builds of master and release builds of the maintained series. Released series (3.3, 3.4, 4.0) are built from their release tarballs with the same packaging script applied externally, so no change to the stable branches is required. This also enables re-packaging revisions (`-1`, `-2`) when a bundled dependency such as OpenSSL needs a security update without a new Ruby release. This part is already working too: in addition to master, zips of 3.3.11, 3.4.10 and 4.0.5 have been built and verified this way, and they are the packages the gem survey above was run on. **3. `rbmanager`: a version manager for Windows.** The same shape as Python's install manager (PEP 773, https://peps.python.org/pep-0773/) A single native executable (the `rb` command) installs a zip from a URL or a local path, extracts it under `%LOCALAPPDATA%`, switches versions with an NTFS junction, and needs one user-level `PATH` entry and no elevation. A working prototype written in C# is already complete and verified on a real machine: ``` > rb install ruby-4.1.0-x64-mswin64_140.zip Extracting ruby-4.1.0-x64-mswin64_140 ... Installed ruby-4.1.0-x64-mswin64_140 > ruby -v ruby 4.1.0dev (2026-06-29T22:22:44Z master 2ce6b8e54a) +PRISM [x64-mswin64_140] > rb list ruby-3.3.11-x64-mswin64_140 ruby-3.4.10-x64-mswin64_140 ruby-4.0.5-x64-mswin64_140 * ruby-4.1.0-x64-mswin64_140 > rb use 4.0 Now using ruby-4.0.5-x64-mswin64_140 > ruby -v ruby 4.0.5 (2026-05-20 revision 64336ffd0e) +PRISM [x64-mswin64_140] ``` The example installs from a local zip today. Once the official version index described below is published, `rb install 4.1` will resolve the version, download the zip and verify its checksum by itself. ## What building on the platform vendor's toolchain enables This proposal is complementary to the existing gcc-based distributions and does not compete with them. Some capabilities are decided by third-party vendors rather than by Ruby, because their SDKs assume the MSVC toolchain and ABI, and those become reachable with an official MSVC-built runtime. Machine-learning runtime assets become directly reusable: the prebuilt Windows packages of ONNX Runtime, libtorch and OpenCV are MSVC binaries with C++ ABIs, and CUDA's `nvcc` accepts only `cl.exe` as its host compiler on Windows, so Ruby extensions binding GPU and ML runtimes need an MSVC-built Ruby to link cleanly. PDB symbols connect Ruby to the standard Windows diagnostics stack: WinDbg and Visual Studio can debug a mixed stack of Ruby and vendor DLLs, WER minidumps from end users become analyzable, and publishing official symbol packages becomes possible for the first time. And extension authors can pull C dependencies from the same vcpkg ecosystem that `ruby/ruby` itself already uses through `vcpkg.json`. ## What I ask to decide First, make these official distribution artifacts: announce them on the www.ruby-lang.org download page and host the zips, checksums and a static version index (the equivalent of pymanager's `index-windows.json`) on ruby-lang.org infrastructure. GitHub Releases can serve as the initial hosting during the preview period. Second, sign the published binaries with a project identity (for example SignPath Foundation issues OSS code-signing certificates) rather than an individual certificate. Third, host `rbmanager` as a repository under the `ruby` organization. The concrete zip layout and naming rule are implementation details. I will work them out with @nobu and document the result in `doc/distribution/windows.md`, so they are not something to decide on this ticket. ## Scope and non-goals This proposal does not replace or deprecate any existing third-party distribution. It adds an official channel for the platform the core team itself builds and tests. I already operate the official snap distribution and the entire release pipeline including security releases, and these artifacts ride on the same automation and the same operator, so this proposal does not change the current maintenance level. Pure-Ruby gems work with no toolchain. Building a C extension requires VS Build Tools, which Microsoft has licensed for building open-source dependencies without a paid Visual Studio license since 2022, and a `ridk`-like bootstrap (`win32/install-buildtools.cmd` is the seed) is follow-up work. Precompiled `x64-mswin64` gems across the ecosystem are long-term work and not a prerequisite. The survey above identified 12 gems that already support Windows with `x64-mingw-ucrt` precompiled binaries and only lack a working MSVC source path, a concrete and small upstream fix list. That list includes nokogiri and ffi, so applications depending on them (Rails, for example) do not work on this distribution yet. The fixes are tractable: I have already verified a one-line fix for msgpack locally and will send these upstream. TLS trust anchors are not bundled in the zip. `rbmanager` provisions certificates from the Windows certificate store at install time, and I will separately propose `OpenSSL::X509::Store#add_store` support for the OpenSSL 3.2+ `org.openssl.winstore://` store in ruby/openssl as the permanent solution. x64 comes first. arm64 (`arm64-mswin64`) already builds in CI and will follow once its test suite is green. ## Roadmap The rollout is incremental rather than tied to a single date. Zips for the released series (3.3, 3.4, 4.0) will be published progressively for their existing releases. For 4.1, nightly zips of master are published continuously, and a zip follows each preview release as soon as it is out. Only the GA of `rbmanager` and the distribution mechanism as a whole is aligned with the Ruby 4.1 release in December. -- https://bugs.ruby-lang.org/

1 0

[ruby-core:125796] [Ruby Feature#22121] Introduce Parallel Sweep feature
by luke-gru (Luke Gruber) 07 Jul '26

07 Jul '26

Issue #22121 has been reported by luke-gru (Luke Gruber). ---------------------------------------- Feature #22121: Introduce Parallel Sweep feature https://bugs.ruby-lang.org/issues/22121 * Author: luke-gru (Luke Gruber) * Status: Open ---------------------------------------- ## Abstract Ruby's GC sweep implementation is currently incremental and lazy. This is to reduce pause times when sweeping. However, the sweep implementation doesn't take advantage of parallelism (multi-core). Sweeping in a GC is not an "embarassingly parallel" problem, but extra threads can help. I would like to introduce such a feature to Ruby so that users can take advantage of their multi-core CPUs to reduce GC pause times. ## Design I would like to have an additional "sweep thread" that sweeps alongside the Ruby GC thread (in parallel) and at the same time as Ruby code is running (concurrent). When the sweep thread is working alongside the Ruby GC thread, both threads grab pages from the current heap. If the sweep thread has already swept a page, the Ruby GC thread finishes it by clearing its bitmaps and adding the page to the free pages or empty pages lists. If there are no swept pages from the sweep thread, the Ruby GC thread sweeps a page by itself instead of waiting for swept pages. When an incremental sweep step is over, the sweep thread sweeps 1 incremental step's budget worth of slots while Ruby code is running. This is so that during the next incremental step, the Ruby GC thread just has to finish the pages off instead of sweeping the objects. ## Limitations Certain objects aren't safe to be freed by the sweep thread. T_DATA types from native extensions cannot be swept in general (although most are safe) because the user's sweep function may not be thread-safe. That function may modify global state in such a way that when it is called from both the sweep thread AND the Ruby GC thread at once, it behaves badly. In order to get around this, we introduce a new [TypedData flag](https://github.com/ruby/ruby/pull/17395). T_DATA internal to the VM mostly have this flag set, and native extension authors can set this flag if it is defined to allow their type to be swept concurrently. This feature is only available for the default GC. Ruby's MMTK garbage collector has its own implementation of concurrent sweeping that is not affected by this feature. This feature is only available for pthread platforms, although that restriction could be lifted with a bit of work. This feature is not on by default (see section `Building`). ## Implementation The [PR](https://github.com/ruby/ruby/pull/17411) is currently a draft, but it's in a working state. Please play with it and tell me what you think! ## Building ```bash ./configure --enable-parallel-sweep make -j ./ruby --enable-parallel-sweep -v #=> ruby 4.1.0dev (2026-06-19T15:49:17Z parallel-sweep cd7e59d45b) +PRISM +Parallel-Sweep [arm64-darwin25] ``` ## Benchmarks This image below is from running [ruby-bench](https://github.com/ruby/ruby-bench) `--headline` benchmarks 10 times and taking the median of means of the runs. The ruby-bench command was: ```bash ./run_benchmarks.rb --no-pinning --interleave --chruby="ruby_ctl::ruby_ctl --yjit" --chruby="psweep_lockfree::psweep_lockfree --yjit --enable-parallel-sweep" --headline ``` If you want sweep info per run, you can run it with the GC harness: ```bash ./run_benchmarks.rb --no-pinning --interleave --chruby="ruby_ctl::ruby_ctl --yjit" --chruby="psweep_lockfree::psweep_lockfree --yjit --enable-parallel-sweep" --harness=gc --headline ``` <img style="width: 1219px;" src="clipboard-202606191327-7umxq.png"><br> I encourage others to experiment with this feature, run your own benchmarks and share them here. When running your benchmarks, make sure not to pin the process to a CPU with `taskset`, otherwise the sweep thread will not run in parallel. > [!NOTE] > Sometimes there are regressions run to run. Run the benchmark 5-10 times to get more accurate results. > GC micro-benchmarks may not see any improvements or even slight regressions. The implementation targets a workload where there is sufficient time in between sweep steps for the sweep thread to do its work. > [!WARNING] > Benchmarking on MacOS tends to be less accurate than Linux. However, if that's all you have and want to share your results, please do. ## Future Work In the future, I would like it if we could process the metadata of a page in the sweep thread. For example, if we could clear the bitmaps and unlink an empty page, or even add a page to the free pages list then the Ruby pause time would get even smaller (or none at all). I did have a prototype of this, but it caused issues because it was creating too many empty pages too fast, and that had unintended consequences for the rest of the GC. I also believe parallel marking would have a large benefit. Major GCs take a long time due to marking, and if you have lots of threads of fibers than it takes even longer. Marking is a more naturally parallel problem, so would likely benefit from more than 1 worker thread. ---Files-------------------------------- clipboard-202606191327-7umxq.png (342 KB) -- https://bugs.ruby-lang.org/

2 5