Activity
From 09/23/2026 to 09/29/2026
Today
-
07:02 PM Ruby Revision b74b3b45 (git): [ruby/prism] Add a constant id set
- There are times when we don't need explicit order in a list of
constants, namely when looking at names of captures in a pattern
match and names of captures in a regular expression. In these cases
it's better to be linear with respect to ... -
- The parse.y grammar has the concept of a words separator that must
delimit words in a word list literal. Prism previously didn't emit
these in the token stream because they don't actually correspond
to any source code.
However, in order... -
- A bunch of things are required to match ripper state, including
looking backward sometimes to pull a state forward because of the
nature of parse.y. This changes the structure from a backward look
to a forward look by keeping a couple of... -
05:47 PM Ruby Revision 8fdf4342 (git): ZJIT: Bake block ISEQ pointers as GC offsets (#19108)
- 03:02 PM Ruby Revision 696ebaad (git): JITs: Rebuild libjit.rlib when jit/src/lib.rs changes
- Both libyjit.rlib and libzjit.rlib list jit/src/lib.rs as a dependency
in release builds containing both JITs. Any change to that file causes
the JIT-specific libraries to rebuild, but they'll rebuild with a stale
libjit.rlib, which can ... -
01:06 PM Ruby Revision aeb212ec (git): [DOC] Adding section 'Cleaning Up' to doc guide (#19101)
- * [DOC] Adding section 'Cleaning Up' to doc guide
* Update doc/contributing/documentation_guide.md
Co-authored-by: Jeremy Evans <code@jeremyevans.net>
* Update doc/contributing/documentation_guide.md
Co-authored-by: Jeremy Evans <cod... -
12:14 PM Ruby Bug #22390 (Open): ruby 3.4 -> 4.0 upgrade GC changes causing >2x RSS increase
- Hello, I've been trying to upgrade a fairly sizeable monolith, which has been running on ruby 3.4.10 in production, to ruby 4 (have been targeting 4.0.6 more specifically, although the below should apply the same to 4.0.7). While we didn...
-
11:35 AM Ruby Revision d1296b10 (git): [ruby/rubygems] Add --source option to gem exec command
- Allows specifying a custom gem source when running `gem exec`, matching
the existing --source option on gem install.
https://github.com/ruby/rubygems/commit/6990083d9d
Co-Authored-By: akshay-birajdar-zendesk <akshay.birajdar@zendesk.com> -
11:17 AM Ruby Revision 43dab371 (git): Stop pending test_sending_hook_payloads_under_gc_stress on Windows arm64
- Its failures on 11-arm no longer fail Windows result, and a crash there now leaves a dump to investigate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Test workers on Windows 11-arm die without any report. With RUBY_DEBUG=wer, Windows Error Reporting writes a minidump for such a crash, and it is uploaded with the binaries and PDBs needed to read it. The runner images already turn off t...
-
- SEM_NOGPFAULTERRORBOX keeps a crashing test from waiting on the WER dialog, but it also keeps WER from writing a dump when the crash bypasses Ruby's handler, as in a thread other than the main one. Such a crash leaves no [BUG] report eit...
-
- The 11-arm check job is not stable enough to block merges. The windows-11-arm label moves to the Visual Studio 2026 image by September 30, and on forks already running it, test workers die in the first TLS handshake of every test-all run...
-
11:13 AM Ruby Bug #22388 (Closed): Use-after-free when an Enumerable retains its `each` block
- Applied in changeset commit:git|c861619e891632aeeb42907b17003dbc9fbf5f2f.
----------
[Bug #22388] Fix use-after-free in some `Enumerable` methods -
- Some `Enumerable` methods pass a pointer to stack-allocated state to `rb_block_call`. If an implementation of `each` retains the given block and invokes it after the `Enumerable` method has returned, the callback accesses invalid stack ...
-
10:36 AM Ruby Bug #22389 (Open): Hang on GC.compact
- The following code works on 4.0.7 but hangs on 4.1.0dev (64a4a372d7):
```ruby
ready = Ractor::Port.new
Ractor.new(ready) do |port|
Array.new(10_000) { Object.new }
port << true
Ractor.receive
end
ready.receive
Thread.new... -
-
-
- Applied in changeset commit:git|64f4b9ab83c4da368e6314c7ab8b74247894cada.
----------
[Bug #21142] Add `Enumerator::Lazy#each_with_index` -
-
- Also, make `lazy_use_super_method` hash `compare_by_identity` since
all the keys are symbols, . -
- Closing as a duplicate of #18439. YJIT on mswin is waiting for review at https://github.com/ruby/ruby/pull/19114.
-
- I made YJIT work on mswin x64 in https://github.com/ruby/ruby/pull/19114. It follows the Win64 calling convention and registers unwind info for JIT code, as discussed here, and passes `nmake check` with YJIT enabled. YJIT is about 2x fas...
-
07:26 AM Ruby Revision 64a4a372 (git): [ruby/mmtk] Update mmtk-core and remove LOS workaround
- There was a bug in mmtk-core that caused `is_live()` and
`is_reachable()` to return wrong results for objects in the large object
space. The function `object_survives_current_gc` was introduced to
workaround the bug.
The bug has been f... -
- When stdin is closed, a socket can get fd 0, and nfds stayed 0 if it was the only fd to wait for, so select never reported the connection.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> -
- The fd was closed but left in connection_attempt_fds, so it was passed to the next select and closed again in the cleanup, possibly after the number had been reused.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> -
- remove_connection_attempt_fd shifts the remaining fds down, so advancing the index skipped the fd moved into the current slot. A connection that had already succeeded could go unnoticed until the next select, and a new attempt was starte...
- 07:10 AM Ruby Revision 16d68e7f (git): [ruby/json] JRuby: keep the whole character after an invalid escape
- With allow_invalid_escape, the JRuby decoder appended the code point read
after a stray backslash with append(int), which writes only its low byte.
"\\é" became "\xE9" and "\\あ" became "B", where the C extension
keeps the character. Use ... -
- The main thread polls has_syserr on every loop while the resolver thread may be setting it, so read it with the lock the resolver thread holds. Once it is seen set, err and ai were written before it and can be read as before.
Co-Authore... -
- The resolver thread checks notify and writes to it while holding the lock. Closing it outside the lock let the thread write a byte to an unrelated file that had reused the fd number.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> -
- The millisecond timeout of a Win32 wait is rounded up to the timer tick, so `sleep(0.001)` took 15.6 ms on average. A waitable timer created with `CREATE_WAITABLE_TIMER_HIGH_RESOLUTION`, accepted since Windows 10 1803, brings that to 1.5...
-
- configure.bat already writes INSTALL_STATIC_LIBRARY, but Makefile.sub never substituted it, so tool/rbinstall.rb could not see --disable-install-static-library. Default it to yes, as configure.ac always sets yes or no.
Co-Authored-By: C... - 05:05 AM Ruby Revision 58b42841 (git): Resolve mkdepend source names against the source root
- A relative name was expanded against the current directory first, so a
build directory nested inside the source tree made generated files such
as ext/ripper/ripper.y look like source-tree files, and the extension's
own depend file was no... - 02:42 AM Ruby Revision ef66863a (git): Update bundled gems list as of 2026-09-28
- 02:40 AM Ruby Revision a23db2a9 (git): Bump the github-actions group across 1 directory with 5 updates
- Bumps the github-actions group with 5 updates in the / directory:
| Package | From | To |
| --- | --- | --- |
| [ruby/setup-ruby](https://github.com/ruby/setup-ruby) | `1.325.0` | `1.327.0` |
| [github/codeql-action/init](https://github... -
01:34 AM Ruby Revision 5266c788 (git): Fix out-of-bounds crash in Array#repeated_permutation
- If the array is modified in the call to to_int in Array#repeated_permutation,
it can trigger an out-of-bounds because the length is cached from before
the call to to_int.
The following script demonstrates teh crash:
ary = (1..1000)... -
- Since the Windows-specific definition was removed, STAT is only an alias of stat, which include/ruby/win32.h already maps to rb_w32_ustati128.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
09/28/2026
-
10:53 PM
Ruby
Bug #22387 (Closed): str_shared_replace incorrectly recalculates termlen
- Applied in changeset commit:git|67cd1d3bde0fc3c1afe2fab3aaac8c10edd12e5b.
----------
[Bug #22387] Fix termlen in str_shared_replace (#19104)
[Bug #22387] Fix termlen in str_shared_replace
The buffer is copied over based on the new str... -
08:52 PM Ruby Bug #22387: str_shared_replace incorrectly recalculates termlen
- Backport PR for 3.4: https://github.com/ruby/ruby/pull/19105
-
- bakport for 4.0: https://github.com/ruby/ruby/pull/19104
-
- PR for master: https://github.com/ruby/ruby/pull/19103
-
- This ruby snippet crashes under ASan with a heap-buffer-overflow
```ruby
s = ("\u{30AF}" * 7).encode("UTF-16LE").b
s.force_encoding("UTF-16LE")
s.encode!("UTF-8") # 21 bytes in a 22-byte malloc, capa now claims 22
s << "x" ... - 10:53 PM Ruby Revision 67cd1d3b (git): [Bug #22387] Fix termlen in str_shared_replace (#19104)
- [Bug #22387] Fix termlen in str_shared_replace
The buffer is copied over based on the new string's termlen
so we can just set the new encoding raw
instead of recalculating based on the old encoding's termlen. -
10:00 PM Ruby Feature #22226: Ractor: class/module ownership -- restrict modification to the Ractor that created it
- matz (Yukihiro Matsumoto) wrote in #note-6:
> Top-level class and module names in non-main Ractors, and the integration with `Ruby::Box`, should be settled together with the author of `Ruby::Box`. Let us discuss that separately.
ko1 ... -
- That was quick. Thank you!
-
- Applied in changeset commit:git|13411f495b47d52cd07de667908073bd5fcea5fc.
----------
[Bug #22386] Reject port creation after Ractor termination
Raise `Ractor::ClosedError` to prevent child threads from accessing
freed port tables durin... -
- [GH-19096](https://github.com/ruby/ruby/pull/19096)
-
04:59 PM Ruby Revision 5fe7d61b (git): [ruby/json] Ensure `default_sort_keys_proc` isn't reassigned.
- It's assignable from Ruby simply for the gem's own initialization,
but it should never be re-assigned later.
Fix: https://github.com/ruby/json/pull/1079
https://github.com/ruby/json/commit/cc32ef5cf5 -
04:22 PM Ruby Revision 42bf37a4 (git): [ruby/prism] Only report real captures inside alternation patterns
- When a variable is captured anywhere in a pattern, the operands of each
alternation are visited to report captures inside them. That visitor
treated every local variable target as a capture, including
underscore-prefixed names and target... -
- Make hidden shape copy of an array.
-
-
-
- Set seconds and nanoseconds together to avoid duplicate conversions.
-
- Keep the cleanup path and reject temporary paths that cannot fit.
-
- Allocate only the buffer needed by the selected home directory source.
-
- Avoid reading before the buffer when the resolved base path is empty.
-
- Raise `Ractor::ClosedError` to prevent child threads from accessing
freed port tables during teardown. - 02:05 PM Ruby Revision fb39511f (git): ZJIT: Profile opt_case_dispatch (#19080)
- In 7eab8b5d488da8474d7b31196e3d06b2ff90399d I wanted to avoid extra
work. Since we don't actually use any profile information in ZJIT for
compiling `opt_case_dispatch`, I decided not to call the profiling
instruction.
However, this caus... -
12:33 PM Ruby Misc #22320 (Closed): Build fails with SIGSEGV in `miniruby` during `builtin_binary` on macOS 26 (Tahoe): `pipe2` declared by SDK but absent from libSystem at runtime
- Thanks for the report!
This is a duplicate of #22311 and should already be resolved in the latest ruby 4.0 release (4.0.11). The next 3.4 release will also contain the fix for this. -
- This was implemented in https://bugs.ruby-lang.org/issues/17314
-
- The fix in commit 11f57c6 contained a bug where the calculated index should
be `len - i` and not `len - 1`. This script reproduces the crash:
ary = (1..100).to_a
gen = Object.new
gen.define_singleton_method(:rand) { |lim| 8 ... -
-
- Windows 7 and older could not write more than about 64KB to a console in one call, so console writes and binary writes to character devices were split into 31366-unit pieces [Bug #14942]. Windows older than 8 has not been supported since...
-
- The child dies with SIGSEGV in about a quarter of the Windows 11-arm check runs on master, and leaves no [BUG] report to go on. It has not shown up on x64 Windows, Linux or macOS.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- LONG2FIX does not check the range, so `[1].cycle(2**30).size` returned -1073741824 on mswin. 64-bit platforms hit the same with 2**62.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Gem::Specification#to_ruby now escapes these values or raises for them, so such a spec fails in ensure_loadable_spec as corrupt. Call to_ruby inside its rescue so that failure is reported the same way.
https://github.com/ruby/rubygems/c... -
- Dependency names were written into %q<> without escaping, and specification_version, dependency types and stub line values were interpolated as is, so a stray `\`, `>` or newline changed the structure of the generated gemspec. Dump depen...
- 06:16 AM Ruby Revision 16926919 (git): Make `IO::Buffer` slices offset-based so they survive source relocation. (#19090)
- 04:13 AM Ruby Revision c568687e (git): Treat empty `IO::Buffer` ranges as non-overlapping. (#19088)
-
- Gem::RemoteFetcher and the Bundler downloader resent the header on every redirect, including one to another origin. It carries the source URI that Bundler replaces with a mirror, credentials included, so it is meant for the mirror's orig...
-
- 02:07 AM Ruby Revision 4d56c955 (git): Revalidate `IO::Buffer` byte access after Ruby callbacks. (#19086)
- Treat encoding coercion, `to_io`, and scheduler capability checks as
callouts that can change the buffer before byte access begins.
- Copy `get_string` bytes through the existing locked-for-reading helper
after encoding coercion, with... -
-
-
-
-
09/27/2026
-
- check_misc asks for this after man/ruby.1 changed.
https://github.com/ruby/ruby/actions/runs/36075733323/job/107886392990
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- None of these removals had been announced in NEWS before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- IRIX is not a platform Ruby supports.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- AmigaOS is not a platform Ruby supports.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- AtheOS, AmigaOS, HI-UX/MPP, IRIX, OSF/1, QNX, ESIX, UX/PDS, System V Release 4 and RISC OS are not listed under Platform Maintainers in doc/maintainers.md, and none of these branches is reached on a supported platform.
Co-Authored-By: C... -
- Ruby 1.9.1 dropped Windows CE, and Ruby 2.3 dropped OS/2 (emx) and Borland C.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Ruby 1.9.1 dropped DJGPP and Windows CE, and Ruby 2.3 dropped Borland C. Gem::WIN_PATTERNS itself stays because other gems reference it. msdos in Bundler::WINDOWS goes with djgpp, since only DJGPP's msdosdjgpp host_os ever matched it.
C... -
- Ruby 2.3 dropped Borland C.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Ruby 1.9.1 dropped DJGPP, Windows CE and VMS, NeXTSTEP, OpenStep and Rhapsody went in 2011, and Ruby 2.3 dropped OS/2, BeOS and Borland C. None of these names appears in RUBY_PLATFORM or host_os on a supported platform.
Co-Authored-By: ... -
- Ruby 1.9.1 dropped DJGPP and Windows CE, and Ruby 2.3 dropped Borland C.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Ruby 2.3 dropped Borland C.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
10:12 PM Ruby Bug #22284 (Closed): Passing an object with to_int to curry as its arity causes problems with LLP64.
- Applied in changeset commit:git|ec3b6b12e7fdd9eb833ba867eb862956a85f424b.
----------
[Bug #22284] Fix curry with a non-Fixnum arity on LLP64
FIX2INT does not call to_int where int and long have the same size, so `method(:puts).curry(1r... -
- NUM2INT already takes the Fixnum path by itself.
-
- It also counts the to_int calls, so it fails on LP64 platforms too, where the old code converted the arity again on every call.
-
- FIX2INT does not call to_int where int and long have the same size, so `method(:puts).curry(1r)` waited for a garbage number of arguments on Windows. Convert the arity once and keep the result as a Fixnum, which also limits the to_int ca...
- 09:20 PM Ruby Revision 4c5d52f0 (git): [ruby/rubygems] Bump rubyzip from 3.3.0 to 3.4.0 in /spec/realworld/fixtures/warbler
- Bumps [rubyzip](https://github.com/rubyzip/rubyzip) from 3.3.0 to 3.4.0.
- [Release notes](https://github.com/rubyzip/rubyzip/releases)
- [Changelog](https://github.com/rubyzip/rubyzip/blob/main/Changelog.md)
- [Commits](https://github.c... -
- Ruby 4.1-dev can segfault when a thread inside a terminating Ractor raises an exception in an `ensure` block. Automatic thread exception reporting tries to load error-formatting code through `require`. The require path creates an interna...
-
- A follow-up on "every form I tried gives the same answer": while writing more tests I found a few forms where it's Prism that gets it wrong, the other way around. With a sibling capture present, Prism also rejects `in [x, 1 | _a]`, `in [...
-
- GetVersionEx reports 6.2.9200 on Windows 8.1 and later unless the manifest of the executable declares them as supported, and ruby.manifest does not. RtlGetVersion is not affected by the manifest.
Co-Authored-By: Claude Opus 5.5 <noreply... -
- A missing variable turned the lower bound into 0, which the existing assertion already covers.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> -
- mingw-w64 guards the macro with _WIN32_WINNT >= 0x0601 as well as the function, while the mingw build targets 0x0600.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> -
- GetSystemInfo() reports the logical processors of the current processor group only, so Etc.nprocessors returned at most 64 on Windows machines with more than 64 logical processors. GetActiveProcessorCount(ALL_PROCESSOR_GROUPS) counts eve...
-
- The child aborts with [BUG], and symbolizing the C level backtrace for the crash report can take longer than the default 10 seconds on a loaded macOS CI host. Use the same budget as 1712837b7f.
https://github.com/ruby/ruby/actions/runs/... -
- Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
-
- Breakages specific to i386-mswin32, such as the size_t atomic fetch_add and the prism bit.h build failure, were found only by auditing. Run check on x86 as well so that CI catches them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic... -
- On i386-mswin32 the ruby DLL has no arch prefix and is named vcruntime140-ruby410.dll, so the vcruntime140*.dll check rejected every x86 binary package. Match only the VC runtime DLL names.
Co-Authored-By: Claude Opus 5.5 <noreply@anthr... -
- On 64-bit Windows, WOW64 sets PROCESSOR_ARCHITEW6432 when a 32-bit process starts, like PROCESSOR_ARCHITECTURE under the x64 emulation on ARM64, so test_execopts_env and test_execopts_unsetenv_others fail on i386-mswin32.
Co-Authored-By... -
- `rb_w32_write_console` returns the number of UTF-16 code units it wrote, so writing "あ" to a console returned 1 instead of 3 as it does for files and NUL. Use the result only to tell success, as the other callers do.
Co-Authored-By: Cla...
09/26/2026
-
- ruby_push_include was its last user. A path separator never appears inside a multibyte character, so walk RUBYLIB and -I by bytes everywhere. This also fixes the same skipped separator on Cygwin, where CharNext came from windows.h.
Co-A... -
- The dln_find.c object built on its own receives UTF-8 command names from process.c, but scanned them with CharNext in the ANSI code page. With a DBCS code page, a backslash after a multibyte character was missed, and `exec("あ\\foo", "x")...
-
- RUBYLIB and -I are UTF-8 on Windows, but ruby_push_include walked them with CharNext in the ANSI code page. With a DBCS code page such as 932, the last byte of a UTF-8 character was taken as a lead byte and swallowed the following `;`, s...
-
- The polling loop restored every socket into the sets before checking
the deadline, so a timed-out select returned them all as ready. When
IO.select forced a zero timeout for buffered data, sockets with nothing
to read were reported read... -
- When FD_SETSIZE (64) or more sockets were ready, copy_fd stopped there
and a ready pipe or other non-socket handle was left out of the result.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> -
- The polling loop saved and restored the sets through fixed fd_set
buffers, which hold only FD_SETSIZE (64) sockets. From the second
poll on, sockets past that were no longer watched, so IO.select could
miss them until the timeout.
Co-A... -
- Nothing in them has used <fcntl.h> or the L_* and *_OK macros since 9afa9ba967 split dln_find.c out and f9e9eee677 removed ruby_add_suffix.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Neither mswin nor mingw defines HAVE_GETGROUPS, so the remaining condition already covers Windows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- include/ruby/win32.h already maps stat, lstat, access, truncate, utimensat, unlink, rename and getenv to the same rb_w32_* functions when compiling the core with RUBY_EXPORT.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- It has been `#if 0` since 04e95f8985 disabled it to avoid infinite recursion at loading transcoder.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Timeout.timeout in Ruby 2.3 takes at most two arguments, so EnvUtil.timeout always raised ArgumentError there. EnvUtil.terminate reaches it through Debugger#dump on Windows when gdb is installed, which broke test_terminate_reaps_exited_c...
- 07:24 PM Ruby Revision b58bea07 (git): Fix buffer size accounting for `rb_str_tmp_frozen` roots
- capa does not include the terminator, so when a buffer moves
between strings with different terminator lengths the capacity has to
be converted or we will try to free the buffer with the wrong size:
[BUG] buffer 0x00005030039ff768 f... - 04:34 PM Ruby Revision 25f659ba (git): ZJIT: Don't CheckInterrupts on inlined method return (#19065)
- An inlined `leave` is just a `Jump` to the caller's continuation block, not a real return no frame pops there, so checking interrupts is pointless. Moved `CheckInterrupts` so it only runs on real (Standalone) returns and real returns are...
- 03:38 PM Ruby Revision 7eab8b5d (git): ZJIT: Add null support for opt_case_dispatch (#19040)
- We are not going to optimized switch-based dispatch with the hash table, at
least not yet, but we should let the interpreter profile the execution paths
before ZJIT runs them.
This should inject a lot more type information for the compi... -
-
- Makes it consistent with nodes that contain text, like `StringNode`
For those, `value` returns the literal. But we can't make it return a symbol for backwards compatibility.
https://github.com/ruby/prism/commit/7c1e135205 -
01:35 PM Ruby Bug #22385 (Open): Enumerator::Lazy#zip loses to_ary conversions when mixed with enumerable arguments
- `Enumerator::Lazy#zip` discards a successful `to_ary` conversion when a later argument requires Enumerable iteration. The equivalent eager `Enumerable#zip` call preserves the conversion.
## Reproduction
```ruby
array_like = Object... -
- Currently, if GC is disabled, postmortem GC is not ran when a Ractor dies.
This can cause the object space to have issues because we need to run the
postmortem GC. This commit will guarantee that it runs. -
12:38 PM Ruby Bug #22342: IO#gets can ignore the limit argument in some edge cases
- ruby_3_4 commit:0a33ab92c08c8d1dbba0b3c97781d16b40ac53a8 merged revision(s) commit:227006291d35b2255b3f7a888d75102c129b0472.
-
11:58 AM Ruby Bug #22342 (Closed): IO#gets can ignore the limit argument in some edge cases
- Applied in changeset commit:git|227006291d35b2255b3f7a888d75102c129b0472.
----------
Fix IO#gets ignoring the limit argument
IO#gets first searches for separator candidates by looking for the last
byte of the record separator, and then... -
- [PATCH] Fix IO#gets ignoring the limit argument
IO#gets first searches for separator candidates by looking for the last
byte of the record separator, and then checks whether it is an actual
match.
If a c... -
- IO#gets first searches for separator candidates by looking for the last
byte of the record separator, and then checks whether it is an actual
match.
If a candidate is found at limit-th byte of the stream but is rejected
because it does ... -
- Applied in changeset commit:git|4300f5dda986414b3ddc9957553315e6437c7f1b.
----------
[Feature #21277] Run check on Windows 11-arm
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- On Windows, Process.kill raises Errno::ESRCH for a child that has exited but has not been reaped, so terminate returned a stale $? and left the child to ZombieHunter. TestIO#test_write_32bit_boundary hit this on Windows 11-arm, where the...
-
- mswin takes the CPU name from MSVC, as it reports x64 rather than x86_64 on x64.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
-
- 11:11 AM Ruby Revision 99d1ece9 (git): io: preserve wide character boundaries in raw reads
-
- include/ruby/win32.h and win32/dir.h have mapped them to the rb_w32_u* functions since 5b98b2ce39.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- On Windows the default case already sets hasalpha for '~' because IS_WIN32 is 1.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- DOSISH is defined only on _WIN32, which has used USE_NAME_ON_FS_REAL_BASENAME since 45df1c24d2, so this code is never compiled. It even passes an undeclared `basename` to do_opendir.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
-
- `size_t` may be less than `off_t`
- 10:11 AM Ruby Revision 8094e4e2 (git): Don't lose consumed bytes when IO#read times out (#19053)
- * io: preserve data after read timeout
* io: make timeout regression tests robust against small send buffers
A synchronous write of the whole payload blocks forever once the socket
send buffer fills (8 KiB by default on macOS and Windo... -
- Windows Server 2016 and Windows 10 Enterprise LTSB 2016 reach end of support by January 2027. Version 1809 is the oldest release still supported afterwards, as Windows Server 2019 and Windows 10 Enterprise LTSC 2019 until January 2029.
... -
- Update the default `NTVER` so that the declarations of APIs introduced in Windows 10 are enabled. `_WIN32_WINNT` has no value for version 1809, the new minimum, so `_WIN32_WINNT_WIN10` is the closest one.
Co-Authored-By: Claude Opus 5.5... -
04:24 AM Ruby Revision da3e1fcb (git): Fix init_is_float bug
-
- Redefining the method warns under RUBYOPT=-w, which CI uses, and the test expects nothing on stderr. Prepending a module leaves the original method in place.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- On Windows, File.unlink raises Errno::EACCES while another handle still holds the file open. Tempfile#unlink already ignores that, but FinalizerManager#call only rescued ENOENT, so the exception escaped from the GC finalizer and was repo...
-
- malloc.h, which win32.h includes, declares it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- It was for Visual C++ 2012 and older, which had no inttypes.h.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- The fallback definitions were for Visual C++ 2008 and older, which had no stdint.h.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- The last use went away in cdb9893c55.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Every `RBIMPL_COMPILER_SINCE(MSVC, ...)` condition here holds for `_MSC_VER` 1915, the oldest version the headers accept.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Building ruby needs `_MSC_VER` 1915 or later, but the generated config.h accepts 1910 and later for such a build, so an extension could still be compiled by 1910 through 1914.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
-
- Only the Cygwin branch in localeinit.c calls nl_langinfo_codeset, so on mswin the object just added an unused export to the DLL.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- MSVC 1915 or later is required and Makefile.sub rejects RT_VER below 140, so the values made for msvcrt and msvcrNN only led to that error.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- PROCESSOR_FLAG lost its only definition in cdb9893c55, so ARCH_FLAG has been empty unless given explicitly.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- JRuby on Windows loads its bundled `win32/resolv`, which has no `tcpip_params`, so two of these tests raised NoMethodError there.
https://github.com/ruby/resolv/commit/971a36ee72
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
09/25/2026
- 11:59 PM Ruby Revision c41b67d6 (git): ZJIT: Specialize caller splats with polymorphic lengths (GH-18640)
- Caller splats with multiple profiled lengths currently fall back to
dynamic sends, even when each length could use SendDirect.
For opt_send_without_block, dispatch on profiled splat lengths within
each ISEQ receiver arm and specialize a... -
- `Win32::Resolve` is never defined, so these tests were not run anywhere.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- GCC 15 in C23 mode reads `ANYARGS` as `(void)`, so the bare function pointers are rejected against Ruby headers that do not add the cast themselves, as happened with Ruby 3.1 on Windows.
https://github.com/ruby/resolv/issues/92
Co-Auth... - 07:17 PM Ruby Revision b49347c2 (git): [DOC] Add GC.start global option
- This option defaults to `true` like all other `GC.start` options. It only has
effect for the default GC and when there are multiple running Ractors. If
there's only a single running Ractor, the option is ignored.
A global GC runs a stop... -
06:53 PM Ruby Bug #22381 (Closed): MatchData assigned to $~ is incorrectly reused
- Applied in changeset commit:git|ea3cd7efe1f6bab78572b31774462e891ef01243.
----------
Fix reuse of MatchData assigned to $~
$~ = m = /a/.match("a").dup
/b/ =~ "b"
p m
[Bug #22381] -
- $~ = m = /a/.match("a").dup
/b/ =~ "b"
p m
[Bug #22381] -
06:53 PM
Ruby
Bug #22378 (Closed): `Ractor::IsolationError` in code not using Ractors
- Applied in changeset commit:git|bec5e51c85ded96969fd070ac5c3908d9f544723.
----------
Fix Module/Class#dup of frozen object
This code should not raise an error:
```
@var = Class.new.freeze.dup # unfrozen copy
@var.instance_variable_set(... - 06:53 PM Ruby Revision bec5e51c (git): Fix Module/Class#dup of frozen object
- This code should not raise an error:
```
@var = Class.new.freeze.dup # unfrozen copy
@var.instance_variable_set(:@test, 123)
```
It was raising `Ractor::IsolationError` because of a bug in `rb_fields_tbl_copy`.
This patch also fixes th... -
06:24 PM Ruby Feature #22277 (Closed): Make `rb_gc_register_address` Ractor-local
- Applied in changeset commit:git|2ad6b99a98058ce25de2d1d536f291b93d893834.
----------
GC: scope rb_gc_register_address to the calling Ractor
The registered_globals list made every Ractor's GC scan every slot under
a shared lock. This co... -
- Now that registrations are Ractor-scoped, a store of another Ractor's
unshareable object through a registered address is a use-after-free
waiting for the owner's next GC.
This commit introduces a method to catch it in debug and ASAN bui... -
- The registered_globals list made every Ractor's GC scan every slot under
a shared lock. This commit splits it into per-Ractor lists.
We also maintain a global registry that references every Ractor that has
ever registered so that unregi... -
- JSON::State.default_sort_keys_proc is registered at extension load time
so it's owned by the main Ractor.
When we implement the registration table on each Ractor storing another
Ractor's unshareable Proc into that slot leaves a dangling... -
- last_cwd is set on init in the main Ractor, but Dir.pwd attempts to
store into it. When this is done on a child Ractor we violate the
invariant that the owning and registering Ractor must be the same. On a
normal build the use-after-free... - 03:48 PM Ruby Revision 4ac2a34c (git): [Bug #22383] Fix inverted `STR_SHARED` check in `rb_str_tmp_frozen_release`
- 45a2c95d0f7184c9cd64ddd26699af31bea8675d mistakenly rewrote
FL_TEST_RAW(orig, STR_SHARED) && !FL_TEST_RAW(orig, STR_TMPLOCK|RUBY_FL_FREEZE)
as
FL_TEST_RAW(orig, STR_SHARED | STR_TMPLOCK) == STR_TMPLOCK
requiring orig to not b... -
- Nested bounded repeats let the int product tlen * qn->lower overflow, so
the check that keeps the compiler from unrolling a large repeat could pass
for an enormous one, hanging the compile or emitting wrapped jump offsets.
Compare agains... -
- Concatenated large repeats reach a program past INT_MAX from a few hundred
kilobytes of source, and the offsets emitted into it wrap. Cap the size
while the program is emitted, which does not depend on the length pass that
overflows alo... -
03:04 PM
Ruby
Bug #22383 (Closed): Fix inverted `STR_SHARED` check in `rb_str_tmp_frozen_release`
- Applied in changeset commit:git|57213d44ce7b1a31fc9648e9cd5eb0c4507f4a49.
----------
[Bug #22383] Fix inverted `STR_SHARED` check in `rb_str_tmp_frozen_release`
45a2c95d0f7184c9cd64ddd26699af31bea8675d mistakenly rewrote
FL_TEST_R... -
- Sorry, confusing with [Bug #22382].
-
- > With class Ractor alias join take end unless Ractor.method_defined?(:join), it reproduced on 3.3 an 3.4.
What reproduces?
I do have a few related fixes coming in another PR but this particular condition is still
else if (FL_... -
- With `class Ractor alias join take end unless Ractor.method_defined?(:join)`, it reproduced on 3.3 an 3.4.
-
- 4.0 Backport PR https://github.com/ruby/ruby/pull/19019
-
- PR https://github.com/ruby/ruby/pull/19018
-
- https://github.com/ruby/ruby/commit/45a2c95d0f7184c9cd64ddd26699af31bea8675d#diff-430d86fdb6c4a558ab0f1b6648bbfae1720e8bde84f026e95a52740014752040R1914
mistakenly rewrote
FL_TEST_RAW(orig, STR_SHARED) && !FL_TEST_RAW(orig, STR_... - 03:03 PM Ruby Revision 57213d44 (git): [Bug #22383] Fix inverted `STR_SHARED` check in `rb_str_tmp_frozen_release`
- 45a2c95d0f7184c9cd64ddd26699af31bea8675d mistakenly rewrote
FL_TEST_RAW(orig, STR_SHARED) && !FL_TEST_RAW(orig, STR_TMPLOCK|RUBY_FL_FREEZE)
as
FL_TEST_RAW(orig, STR_SHARED | STR_TMPLOCK) == STR_TMPLOCK
requiring orig to not b... -
- `stack_maxsize` is unknown on wasm and set to 0.
-
-
-
-
-
- Deep native stacks can overrun the caller's buffer while reporting
a crash because the Darwin implementation ignores `size`. -
- Unoptimized ASan VM frames can exhaust the C stack before
`SystemStackError` is raised with the normal reserve. -
- If the format string is modified during Array#pack (by conversion such
as to_str, to_int, etc.), it can cause an use-after-free since reading
p[-1] may be an use-after-free. We should read that byte before performing
the conversion.
The... -
07:09 AM Ruby Misc #22275: Ruby::Box support plan for RubyGems and Bundler
- I think that RubyGems / Bundler should not be loaded into Ruby::Box by default. I have shared a few of my thoughts with @tagomoris and based on that he implemented #22329.
But I'd like to go further. I think that it would be nice to h... -
- _BitScanForward64 is not available for 32-bit x86, so the i386-mswin32 build failed at prism.c. Use the portable fallback there, as ntz_int64 in internal/bits.h does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- USE_SPAWNV needs spawnv without a working fork, which only Windows has since OS/2 support was removed. The non-Windows proc_spawn_cmd_internal and proc_spawn_sh were never compiled, and the Windows proc_spawn_cmd_internal macro was unuse...
-
- Neither configure.ac nor win32/Makefile.sub defines HAVE_SPAWNVE.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- HAVE_FDATASYNC is never defined on Windows. win32/Makefile.sub does not define it and the mingw CRT has no fdatasync, so the branch has never been compiled.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- __BOW__ (BSD on Windows) appears nowhere else in the tree.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- On Windows, ruby/win32.h already includes it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- win32/Makefile.sub and the mingw part of configure.ac already define HAVE_FCNTL_H, HAVE_SPAWNV and HAVE_TIMES, and leave HAVE_SYS_IOCTL_H undefined.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Regexp fixes are merged into ruby/ruby directly and the copies have diverged, so asking contributors to file them to k-takata/Onigmo only misleads them.
https://github.com/ruby/ruby/pull/19026
Co-Authored-By: Claude Opus 5.5 <noreply@a... -
- The Bundler downloader, Gem::RemoteFetcher and the compact index fetcher re-applied the source's userinfo whenever a redirect stayed on the same host, even when it moved to another port or scheme. RFC 9110 defines the protection space of...
-
05:21 AM Ruby Bug #22384: Regexp compile error in the 2nd+ branch of a top-level alternation leaks the parse tree (regression from #13332, Ruby 4.0)
- Fix and regression test: https://github.com/ruby/ruby/pull/19026. Verified on a build of v4.0.7 with the patch: Regexp.new("a|b(c") rescue nil goes from 134 B leaked per compile to 0, and 2M calls of /(?:a|b)c(d)/.to_s from +250 MB RSS t...
-
- **Affects:** ruby 4.0.6 and 4.0.7 (official docker images and a jemalloc-linked build), master (commit still present). **Not affected:** 3.4.6.
**Repro (plain Ruby, no extensions):**
```ruby
def rss_kb = File.read('/proc/self/status... -
- GetVersionExW always reports VER_PLATFORM_WIN32_NT, since Windows 9x support had been dropped in 8a02eed723 before Etc.uname was implemented on Windows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- GetSystemInfo reports PROCESSOR_ARCHITECTURE_ARM64 there, which fell through to "unknown".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Thank you for this bug report. I have a [fix here](https://github.com/ruby/ruby/pull/19023).
-
- Otherwise configure succeeds and the same floor in Makefile.sub stops the first nmake.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
-
- Since 87435fe232 the build has needed the conforming preprocessor, which MSVC gained in 1915, so 1900 through 1914 passed this check and failed later with an unrelated message.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Concatenated large repeats reach a program past INT_MAX from a few hundred
kilobytes of source, and the offsets emitted into it wrap. Cap the size
while the program is emitted, which does not depend on the length pass that
overflows alo... -
- Nested bounded repeats let the int product tlen * qn->lower overflow, so
the check that keeps the compiler from unrolling a large repeat could pass
for an enormous one, hanging the compile or emitting wrapped jump offsets.
Compare agains... -
- I believe there is a bug in the fix, when a block with side effects is passed in. I think this should fix it https://github.com/ruby/ruby/pull/19022
-
- Compiling the setter call could spend the whole 1ms budget on a busy CI runner, which stopped compilation before the next eval and failed with :not_compiled2. Check that compilation continues under a generous budget first, and only then ...
- 02:33 AM Ruby Revision 5d106710 (git): Bump taiki-e/install-action
- Bumps the github-actions group with 1 update in the / directory: [taiki-e/install-action](https://github.com/taiki-e/install-action).
Updates `taiki-e/install-action` from 2.87.17 to 2.87.18
- [Release notes](https://github.com/taiki-e... -
- The lockfile does not record overrides, so unlocking for a required_ruby_version or required_rubygems_version override happened on every run, which rejected every frozen install and updated the named gem on each plain `bundle install`. I...
-
- A transitive override outside the parent's requirement made check_lockfile report the parent as having invalid dependencies, so frozen installs of such a lockfile always failed.
https://github.com/ruby/rubygems/issues/9907
https://gith... -
- converge_overrides_outside_dependencies marked every transitive-only version override as a changed dependency, so frozen installs always failed and a plain `bundle install` re-resolved the target as if it had been updated. Compare the lo...
-
- The specification policy looks extensions up through builder_for, so gem build with an extension no builder handles went through build_error. It crashed with a NameError because fileutils was not loaded yet, or wrote gem_make.out into a ...
-
- A reinstall of the same version without build arguments picked up the stale file, so the arguments of the uninstalled copy came back through Gem::Specification#build_args and were reused by gem pristine.
https://github.com/ruby/rubygems... - 12:50 AM Ruby Revision da155f30 (git): [ruby/rubygems] Identify ML-DSA by OID and gate tests by capability
- RubyGems currently assumes an SSL library either supports every ML-DSA
operation or none of them. It also identifies ML-DSA keys using
library-specific display names.
Identify ML-DSA keys using standard SubjectPublicKeyInfo OIDs. Separa... -
- Since a094c99b74 the macro only switches isADS, and every use of isADS is inside `#if USE_NTFS`, so its non-NTFS definition was never referenced.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
09/24/2026
-
-
-
11:57 PM Ruby Bug #22382 (Open): IO#write of a shareable String from multiple Ractors causes use-after-free
- When several Ractors `IO#write` the same shareable, heap-allocated (non-embedded) String at the same moment, Ruby frees the String's buffer while the String is still alive. A later GC sweep double-frees it and aborts.
**Reproduction**... -
- It used InterlockedExchangeAdd64 on a 4-byte size_t, so the access also covered the following 4 bytes, which is the `tdata_unsafe_free_published` pointer for the caller in gc/default/default.c. Go through rb_atomic_t like the other size_...
-
- test_os_code only compared the written os_code with Zlib::OS_CODE itself, so it could not catch OS_CODE falling back to OS_UNIX.
https://github.com/ruby/zlib/commit/104cbd525f
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- mkmf's create_header ignores $defs entries that do not start with -D, so the leading space dropped OS_CODE from extconf.h. The in-tree build of ruby/ruby then fell back to OS_UNIX in zlib.c, and Zlib::OS_CODE was 3 on Windows.
https://g... - 11:34 PM Ruby Revision 5cd1287f (git): Prefer __has_include over HAVE_STDCKDINT_H for <stdckdint.h>
- HAVE_STDCKDINT_H records what configure's compiler saw when ruby was
built. __has_include asks the compiler that is compiling the header
right now. When the two compilers are the same they always agree, so
checking __has_include first ... -
- They were forwarded to `msvcrt.stricmp` and `msvcrt.strnicmp`, which msvcrt.dll does not export, so they could never be resolved. Extensions get `_stricmp` and `_strnicmp` through the macros in include/ruby/win32.h.
Co-Authored-By: Clau... -
- For strings larger than or equal to IO_BUFFER_BLOCKING_SIZE (which is 1KB),
IO::Buffer#set_string will perform the memmove without GVL. In this time,
another thread can resize the IO::Buffer object. In that case, the buffer
may be freed ... -
- Applied in changeset commit:git|c29739cb3ef483c215b0c13d30c96cc21b0f680b.
----------
Implement RubyVM::YJIT.max_compile_time_ns
[Feature #22236]
Allows to give a time budget to YJIT. When YJIT goes over the budget
it pauses compilatio... -
- [Feature #22236]
Allows to give a time budget to YJIT. When YJIT goes over the budget
it pauses compilation until the budget is raised.
It is best effort, YJIT may go over, or have no choice but to compile
in some cases, but the goal i... - 08:46 PM Ruby Revision 03321869 (git): ZJIT: Don't allocate strings for backend labels (#19004)
- There is no sense in eagerly allocating a string; we either use a static
string ref or label a block by some combination of (HIR,LIR) pair. Just
store the information instead of allocating on the heap.
This is probably a very minor comp... -
- ```ruby
m = /a/.match("a").dup # dup a MatchData (has no "busy" flag set)
$~ = m # assign to $~
/b/ =~ "b" # perform a match that sets $~
p m
# "b" on Ruby 4.0 (and 1.8-2.7, incorrect)
# "a" on Ruby 3.0-3.4 (correct)
```
This i... - 06:19 PM Ruby Revision 1533c109 (git): Preserve compare_by_identity in some Set operations
- Backport of f069efc3f69a89188bab4ac70c690aecce59023b
Fixes [Bug #22174]
Co-authored-by: Jeremy Evans <code@jeremyevans.net> -
06:01 PM Ruby Revision ff03ecf3 (git): ZJIT: Print `PatchPoint` reference to `Insn::Snapshot` when printing snapshots
- So both the definition and the usage are shown. Note that this
is only when, say, `--zjit-dump-hir=all`, where both ends are present. - 05:54 PM Ruby Revision 24be314b (git): Enable incremental marking when the process has > 1 objspace
- Previously, major GC incremental marking was disabled (even on the main Ractor)
when more than 1 Ractor was running. By changing the WB implementation, we now
have it back. - 05:17 PM Ruby Revision 4133daa1 (git): Ractor: wake receiver waiters inside the send critical section
- ractor_send_basket used to enqueue the basket under the receiver's lock,
release it, and then re-take the same lock inside ractor_wakeup_all to wake
the receiver. Every sender to a Ractor contends on that one lock, so the
second acquisit... - 04:59 PM Ruby Revision f9b763af (git): ZJIT: Reduce the amount of memory required for profiles (#19000)
- `resize` might over-reserve space speculatively but we know we're only
going to see a certain number of operands at that opcode. So don't
over-reserve; allocate a vec of the exact size we want.
For a run of lobsters, takes `zjit_alloc_b... - 04:58 PM Ruby Revision 10cc9824 (git): ZJIT: Remove `chase_insn` in `fold_constants` for `StringEqual` (#18998)
- Remove `chase_insn` in `fold_constants` for `StringEqual`
`left` and `right` are already canonical operand IDs when `fold_constants`
checks them.
`canonicalize` runs immediately before `fold_constants`.It updates
operands when the opti... - 02:29 PM Ruby Revision 3676b957 (git): ZJIT: Add line number mappings in Comment (#18752)
- Add line number comments to HIR printouts if enabled. Trying this out. Might help for something like Compiler Explorer. Also helps for debugging.
Requires `--zjit-dump-hir-map` to avoid messing with the default printout.
```
plum% cat ... -
02:22 PM Ruby Revision 645a0718 (git): Increase sleep in fork bootstrap test
- There have been occasional failures on the OpenBSD CI machine in
this bootstrap test. There is a race condition in the test that
you can theoretically hit on a very slow machine, and the OpenBSD
CI machine is running in qemu software emu... -
- This test is flaky on OpenBSD RubyCI, probably due to the fact that
the OpenBSD RubyCI machine is a software emulated virtual machine
running on another virtual machine. The test runs fine on OpenBSD
in general, and seems to pass the maj... -
-
- Drop `--repeat-count=2` and its dedicated settings because parallel
runs add CI time without testing state carried across repetitions. -
- Ignore `--repeat-count` during parallel tests because fresh workers
cannot expose state carried over between repetitions. -
- Keep the jobserver connection until all repetitions finish so that
later runs do not launch up to 256 workers without token limits. -
- Run `GC.start` in the process executing tests because collecting the
parent heap does not reclaim objects in parallel workers. -
-
-
-
-
11:33 AM Ruby Feature #22304 (Closed): Add rb_warn_to_remove_at() for deprecation warnings shown by default
- Applied in changeset commit:git|611bb146b08978aacb26bf4a4e8390e699628350.
----------
[Feature #22304] Add rb_warn_to_remove_at and rb_warn_scheduled_deprecation
rb_warn_deprecated_to_remove_at warns only when Warning[:deprecated]
is en... -
- rb_warn_deprecated_to_remove_at warns only when Warning[:deprecated]
is enabled, so a deprecation scheduled to warn by default in a later
version needed a hand-written rb_warn and a separate commit to switch.
rb_warn_to_remove_at emits ... -
-
-
-
-
-
-
-
- rb_file_expand_path_internal is compiled only without _WIN32, where USE_NTFS has been 0 since 7eb8d74f34 stopped enabling it on Cygwin [Bug #13008]. Forcing it on no longer even compiles, as the block still calls strrdirsep with three ar...
-
- ntfs_tail was the only path helper still deriving it from enc on its own after 6fb50434e3 made the callers decide.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Clear temporary containers after extracting the result to avoid
retaining their contents through a captured block. -
- Keep the state for block-based `Enumerable#uniq` alive after it returns,
as `each` can retain and later invoke the internal block, making the
stack-allocated memo introduced in 29c669bb568 invalid. -
- Applied in changeset commit:git|ea2836f7ec905e4963b49ffdc550cdf3e82ad8c9.
----------
[Feature #17548] Allow `Dir.glob` to follow symlinks with `***`
Allow recursive matching through directory symlinks while stopping
at links to ancesto... -
- Allow recursive matching through directory symlinks while stopping
at links to ancestor directories to avoid cycles. -
- I merged the following four, since they fix plain bugs such as crashes and need no design decision.
* https://github.com/ruby/ruby/pull/18704: `alias $new $old` in a box created a separate variable.
* https://github.com/ruby/ruby/pull/1... -
-
- This is a backport request for the following crash:
```
$ RUBY_BOX=1 ruby --disable-gems -W:no-experimental -e 'GC.stress = true; module Enumerable; def m; end; end'
-e:1: [BUG] try to mark T_NONE object (obj: 0x00000001042efe28 T_N... -
- When the string buffer reaches BUFSIZ (8192 bytes), Marshal.dump will call
the write method on the IO object. This IO object can run arbitrary Ruby
code. If the object is a string, it writes the length of the string followed
by the bytes... -
08:19 AM Ruby Revision cc32a303 (git): [ruby/error_highlight] Use Thread::Backtrace::Location#syntax_tree when available
- https://github.com/ruby/error_highlight/commit/7755b9bff1
-
- Under RUBY_BOX=1 an assignment in a user box, the main box included, only
updates the box-local table and never reaches the global entry, so hooks
registered with trace_var never run.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> -
-
-
- The Enumerable version did not crash an unfixed build on Linux. Kernel
does on both Linux and macOS, but only while nothing in the main box has
written to it yet, and loading test/unit already has. So run the code
through assert_in_out... -
- rb_class_duplicate_classext builds the whole copy first and the caller only
then puts it in the class's table, so a GC in the middle collects the tables
it has made and marking the class afterwards walks into a freed object. Put
the cla... -
- rb_gvar_set and rb_gvar_get key the box table by var->id, so defined? on an aliased name looked for a key that is never written and returned nil.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> -
- Under RUBY_BOX=1 the box-local gvar table is keyed by the name used for
access, so `alias $b $a` gives the two names separate slots and they
stop being synonyms. Key it by the ID the shared rb_global_variable
was created with instead.
... -
- #21795 added `#syntax_tree` for a few objects. Currently it is implemented in a way that uses the prism gem, using node_id. Because it is theoretically possible that node_ids from the internal cruby version of prism don't match with the ...
-
- Clear the OpenSSL error queue when an OpenSSL function fails with
SSL_ERROR_SYSCALL. ruby/openssl methods are expected not to leave stale
entries in the error queue.
While this no longer appears to happen with OpenSSL >= 3.0, LibreSSL
a... -
- It was only used to skip the Rust `cargo test` example while magnus could not build against Ruby master, and that skip went away with the magnus 0.9 template update.
https://github.com/ruby/rubygems/pull/9904
https://github.com/ruby/ru... -
06:02 AM Ruby Bug #22174: Set operations (&, ^, collect!, flatten, classify, divide) do not preserve compare_by_identity
- 4.0 backport PR: https://github.com/ruby/ruby/pull/18992
-
- Fixed in commit:f069efc3f69a89188bab4ac70c690aecce59023b
-
- File.expand_path walks past the root through its own call site, and
Pathname#sub_ext skips the length check File.extname does first, so both
reach the backward scan by routes the extname test does not exercise. -
- The fast path added in 11d29d32d2 skips trailing separators without a
lower bound, so a path made only of separators walks the cursor past the
start of the buffer and keeps reading until it meets a byte that is not
a separator. The retur... -
- Oddly, when building with msvc (19.44), the linking fails only while disabling optimization (using the /Od option).
```
linking miniruby.exe
Creating library miniruby.lib and object miniruby.exp
set.obj : error LNK2019: unresolved ext... - 04:48 AM Ruby Revision a3c583af (git): [ruby/rubygems] update magnus version in rust extension gem template
- https://github.com/ruby/rubygems/commit/666723ae10
-
- The size assumed that `fd_array` follows a 4-byte `fd_count`, but on x64 SOCKET is 8 bytes and `fd_array` starts at offset 8. Filling the last slot of a set grown to 128, 192, ... sockets wrote 4 bytes past the allocation.
Co-Authored-B... -
- Apply suggestion from @nobu
Co-authored-by: Nobuyoshi Nakada <nobu.nakada@gmail.com> -
- msvc supports flexible array as it's language extensions.
The /Za msvc compiler option disables this and other extensions, but it is not necessary to consider it.
The Windows SDK headers contain anonymous unions/structs, and using /Za ma... -
- Using the option (one of -std:[c11|c17|clatest]) to enable standard C support in msvc causes the following error:
[BUG] vm_call_method: unsupported method type (-7)
Simply disable it. -
- It should delete `ruby/prism` now
- 03:01 AM Ruby Revision f069efc3 (git): Preserve compare_by_identity in some Set operations
- Adjust the specs to handle the backwards compatibility breakage.
While we will backport this to Ruby 4.0, to avoid temporary
breakage, leave Ruby 4.0 behavior unspecified until the fixes
are backported to it.
Co-authored-by: Jeremy Evan... - 02:30 AM Ruby Revision f55f1957 (git): Bump ruby/setup-ruby in the github-actions group across 1 directory
- Bumps the github-actions group with 1 update in the / directory: [ruby/setup-ruby](https://github.com/ruby/setup-ruby).
Updates `ruby/setup-ruby` from 1.324.0 to 1.325.0
- [Release notes](https://github.com/ruby/setup-ruby/releases)
- ... -
- Applied in changeset commit:git|4617d60fe4b5c18e7350d37b2f2c4508c62df708.
----------
[Bug #22339] Skip classes already freed while freeing a box
`RUBY_FREE_AT_EXIT=1` frees every object in an arbitrary order, so a
class a box refers to... -
- I was already aware of this, and https://github.com/ruby/ruby/pull/18710 fixes it by skipping classes that were already freed when a box is freed.
The ASAN double free in the description comes from `ruby_4_0` and has a separate cause. I... -
- `RUBY_FREE_AT_EXIT=1` frees every object in an arbitrary order, so a
class a box refers to can go before the box itself, and freeing a class
already frees its classext for that box. The box then reached rb_bug
with a T_NONE object, and ... -
- 12:06 AM Ruby Revision 634de041 (git): [ruby/strscan] Don't match from the middle of a character
- (https://github.com/ruby/strscan/pull/219)
`#pos=` accepts an arbitrary byte offset, so the scan position can fall
in the middle of a multi-byte character such as UTF-16BE. Matching from
there made Onigmo read a whole character starting...
09/23/2026
- 10:30 PM Ruby Revision d7bd382a (git): ZJIT: Inline String#to_sym to StringIntern (#18981)
- Closes https://github.com/Shopify/ruby/issues/1061.
Also add a test to make sure `String#intern` generates the same HIR.
### Benchmarks
I had AI create 4 new microbenchmarks for this change and also run `protoboeuf-encode`.
They look... -
-
- Previously, `setlocal level=0` did not set the local in the frame state
during HIR build, unlike `setlocal_WC_0`. `getlocal` and `getlocal_WC_0`
had the same issue.
I don't know if it's possible to trigger an ZJIT issue with default
ise... -
-
- There were some difficulties after https://github.com/ruby/prism/pull/4223 was merged because
the excludes were out of sync. - 09:56 PM Ruby Revision 2ea36fd6 (git): ZJIT: Specialize sends with a Proc block arg into SendDirect (GH-18180)
- Guard the block arg of `foo(&blk)` when profiling shows it is exactly
a Proc, and pass it directly as the callee frame's block handler
instead of falling back to a dynamic send. - 08:36 PM Ruby Revision 9635fc03 (git): Don't allow writing to locals in cexpr (#18873)
- The JIT optimizer sees an `invokebuiltin` and would like to assume that calling it does not write back to the frame. A `Primitive` should never write to the frame by taking a pointer to a local, so this PR removes that functionality.
Pa... -
- 06:40 PM Ruby Revision d8177d64 (git): ZJIT: Move string encoding flags loading to HIR (#18894)
- Closes https://github.com/Shopify/ruby/issues/1051
The LIR generated without stats is the same, but now the `LoadField` can be deduped.
Same as #18251 but without moving the `xor` to the HIR.
Generate HIR is now:
```
v31:CU... -
-
- > Not an area of ruby I'm very familiar with, but I think perhaps we should also be extending sp in vm_callee_setup_arg and vm_call_iseq_setup_kwparm_kwarg?
These don't extend SP because they're fast paths that try to do minimal work... -
- ruby_3_4 commit:41e214d7100100a43c469e2628403d6916d1f0f8 merged revision(s) commit:d87ffb271a80fb108de3465d12ada829f3c130af.
-
04:47 PM Ruby Revision d74c656b (git): Document Range attributes begin and end
-
- Keyword tracking only needs membership, so replace identity hashes
with identity sets to avoid storing unused values. Add `rb_set_p`
for the type assertion in `vm_check_keyword`.
Add benchmarks for 31, 32 and 128 keyword parameters wit... -
- ```rb
# Reduced from the `httpx` gem
@var = Class.new.freeze.dup
@var.instance_variable_set(:@test, 123)
```
I expect this code to succeed. dup returns an unfrozen copy, so I should be able to set instance variables on it. However... -
10:12 AM Ruby Revision 94285c5c (git): [DOC] `Encoding.locale_charmap` never returns nil (#18943)
-
- https://github.com/ruby/mmtk/commit/2801d95aa4
-
09:17 AM Ruby Revision 420a1dee (git): [DOC] Add no argument example to `module_function`'s doc (#18962)
-
08:20 AM Ruby Bug #22254: [Prism] Clang UBSan reports NULL/0 memcpy while serializing empty strings
- Thank you very much for this.
-
- Fixed by https://github.com/ruby/ruby/commit/8cdc7b59571dd5ea4fc08638e88255d3f43d2e60
-
- storage behind `PM_STRING_EMPTY`
(https://github.com/ruby/prism/pull/4228)
Some places memcpy the source, and NULL there is undefined behaviour.
Fixes https://bugs.ruby-lang.org/issues/22254
https://github.com/ruby/prism/commit/d0b36c... -
- If the source string is modified in the block of String#unpack, it can
cause an use-after-free. For example, the following script crashes:
fmt = "A" * 1_000_000
str = "C" * 1_000_000
str.unpack(fmt) { fmt.clear } -
- Applied in changeset commit:git|0fab30f723810af1608d2162e2aef32b44a1ee3a.
----------
compile.c: ensure swapped element isn't skipped
[Bug #22377]
When we statically resolve a `swap` we need to make sure the element
we reordered will r... -
- Thanks for the repro, I think I know what's going on, I'll look into it.
-
- [Bug #22377]
When we statically resolve a `swap` we need to make sure the element
we reordered will receive its optimization pass. -
-
- ruby_3_4 commit:7bbe8e88007f8283a421cb31db86da32076126a6 merged revision(s) commit:ad23fd4a519fde4fc96c7326fb9b0347cbc347b8.
-
- ruby_3_4 commit:75309143a8494abd682bdd23f3b18ecc2e48a2e4 merged revision(s) commit:cb302881629f997f403e705425f69e5f6b0741ac, commit:c914aeb622ce7337223cd5c23f30daf9f7d3e7c4.
-
- ruby_3_4 commit:93d1c2d7c3782cfbe0448849bd2cb79ce66c3860 merged revision(s) commit:ee786598f10c6def72265fb7105fe505f563f52f, commit:54f254aa41eb624429e6b884c0a6381eeeac50d0, commit:15dbb08ba8b69877c8012e0f5950f3f26a3abffe, commit:249de20...
-
- MEMO: related commit hashes.
commit:ee786598f10c6def72265fb7105fe505f563f52f
commit:54f254aa41eb624429e6b884c0a6381eeeac50d0
commit:15dbb08ba8b69877c8012e0f5950f3f26a3abffe
commit:249de200d5be1f1d1f936e40949a6b48ac9b6e25 -
- [PATCH] [Bug] win32: create directory symlinks for relative directory targets
File.symlink decided whether to create a directory symlink by calling
GetFileAttributesW on the target as given, which resolves a relative
... -
- [PATCH] [Bug #22335] Use identity hashes for keyword argument tracking
With more than 31 keyword arguments, tracking unevaluated defaults
uses an integer-keyed Hash whose collisions can invoke methods when
`Integ... -
- [PATCH] [Bug #22340] Fix stat on symbolic-link loops on Windows
Preserve `ELOOP` instead of hiding symbolic-link resolution errors
with the file-enumeration fallback, so existence checks return false. -
- [PATCH] Fix memory leak of transcoding when fallback raises
When the fallback function in transcode_loop raises, it will leak the memory
in rb_econv_t. The following script reproduces the leak:
10.times do
... -
- 02:41 AM Ruby Revision 8c0eb987 (git): Bump taiki-e/install-action
- Bumps the github-actions group with 1 update in the / directory: [taiki-e/install-action](https://github.com/taiki-e/install-action).
Updates `taiki-e/install-action` from 2.87.16 to 2.87.17
- [Release notes](https://github.com/taiki-e... -
- The precision in Time.new calls to_int which can modify the string passed
into Time.new. This can cause an use-after-free. For example, the following
script crashes:
str = "2000-01-01 00:00:00" + "0" * 1_000_000
obj = Object.new... -
-
-