Activity
From 08/24/2026 to 08/30/2026
Today
-
10:04 PM Bug #20243 (Open): M:N threading VM_ASSERT failure in rb_current_execution_context with clang 17 (on Linux)
- Thank you, I understand now.
When I said “the issue,” I was referring to the problem where Ruby crashes when built with Clang 17 and `-DVM_CHECK_MODE=1`.
However, I now realize that this issue does not address that specific case, but r... - 07:29 PM Revision 3aa34522 (git): [ruby/timeout] Exclude the gemspec from packaged files
- https://github.com/ruby/timeout/commit/72a3743ec4
-
05:21 PM Revision aa33db69 (git): Pass the buffer's size to ruby_xrealloc_sized, not the field distance
- rb_econv_insert_output() grew the insertion buffer with
ruby_xrealloc_sized(*buf_start_p, s, buf_end_p - buf_start_p);
The old-size argument subtracts the two local pointer variables instead
of the pointers they point at, so the al... -
- transcode_loop() passed the fallback result to rb_econv_insert_output()
as a bare pointer and length. Nothing roots the string during that call:
args.rep still holds the pre-fallback error-bytes string, and with
clang -O3 the VALUE itse... -
- I simplified the script to make it easier to read in ISEQ.
```
1.times do
begin
redo
ensure
end
ensure
puts 1
break
end
```
```
$ ./miniruby -v --dump=insn a.rb
ruby 4.1.0dev (2026-08-30T06:52:45Z :detache... -
10:19 AM Bug #22268: Backport resolv security fixes (CVE-2026-80212, CVE-2026-80213)
- Thank you for creating PR. merged into ruby_3_4 at commit:6deacfb8275bd1583ffa1b8c2b87f25484586a91.
-
10:18 AM Revision 6deacfb8 (git): Bump up resolv-0.7.2 for Ruby 3.4
- Backport of the August 2026 resolv security release.
CVE-2026-80212
CVE-2026-80213
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> - 06:52 AM Revision a889da90 (git): ZJIT: Remove WriteBarrier for immediates
- It doesn't end up generating any code (we catch this in codegen too) but
we might as well not have the instruction lying around in our IR. - 06:52 AM Revision 4068011d (git): ZJIT: Fold BoxBool(Test(cond:BoolExact)) to cond
- This appears when doing e.g. `l & (a == b)` (as in the included new
test) where `TrueClass#&` does `RTEST(RBOOL(rhs))` and then the right
hand side is already known to be returning a `BoolExact` because it's
from `FixnumEq`. - 06:52 AM Revision 6ce41703 (git): ZJIT: Add line numbers to Perfetto traces
- This helps figure out what part of long functions is causing the exit or
fallback or whatever is being traced. - 06:52 AM Revision b6071e1e (git): ZJIT: Specialize TrueClass#&
- It just does `RBOOL(RTEST(other))`.
- 06:52 AM Revision e5cd8f00 (git): ZJIT: Specialize FalseClass#&
- It just returns `false` no matter the other argument.
- 06:52 AM Revision 0f20fd9f (git): ZJIT: Support generating Const::CBool in codegen
- 06:50 AM Revision f20155d5 (git): marshal.c: Make partial objects table conditional
- Only allocate and use the partial objects table when a load proc is active or loaded objects must be frozen.
Co-authored-by: Codex <noreply@openai.com> -
- A blocking read used to cost four epoll_ctl per wait cycle: ADD before
the wait and DEL after the wake, on both ends of an exchange. The DEL
ran on the timer thread, the one serial resource every io readiness
event already passes throug...
08/29/2026
-
01:05 PM Misc #22272: Propose Kevin Menard (@nirvdrum) as a core committer
- +1, Kevin has a lot of experience with String semantics and optimizations, as well as with implementing & JIT-compiling many parts of Ruby, he would be a great committer.
-
06:49 AM Bug #22269 (Closed): `Coverage.line_stub` clobbers already-collected coverage data
- Applied in changeset commit:git|22e4a75d67d1e25deed4dce5b801397713dbdefa.
----------
[Bug #22269] Make coverage_enabled: false effective for compile/compile_file
Use it in Coverage.line_stub so compiling the stub does not overwrite alr... -
- Use it in Coverage.line_stub so compiling the stub does not overwrite already-collected coverage data.
-
05:51 AM Misc #22249: DevMeeting-2026-09-10
- * [Feature #22279] Region (Length / Range) Arguments for String Bit Operations (hasumikin)
* Follow-up to [Feature #22118] (String bit operations, accepted): adds `(offset, length)` and Range overloads to `bit_set` / `bit_clear`... -
- PR URL: https://github.com/ruby/ruby/pull/18562
## What this adds
Region overloads for the methods introduced in #22118, so they act on a contiguous run of bits in a single call:
```
String#bit_set(offset, length, lsb_first: tr... -
05:17 AM Revision fcbdc207 (git): Dynamically derive base_label in iseq
- base_label isn't used very often in iseq and we can dynamically derive it
from the label (of the current iseq or the local_iseq). We can remove this
field to save 8 bytes. -
03:15 AM Revision e947cbd1 (git): Regenerate build dependencies after revision changes
-
01:04 AM Revision a7361d3e (git): [DOC] Harmonize chmod methods doc
-
08/28/2026
-
11:19 PM Bug #22246: Rational#to_f incorrectly rounds when the numerator is a bignum that cannot be represented exactly by a double, and the denominator is a fixnum
- I found two more cases where `rb_big_fdiv_double` produces an incorrectly rounded result. I added them to [pull request #18599](https://github.com/ruby/ruby/pull/18559).
```Ruby
assert_equal(3.333333333333333e39, (10**40).fdiv(3), "Bug... -
- Windows carried a second, much weaker scheduler: a single Win32 mutex as
the GVL, vm->clock for preemption, its own VM barrier over
vm->ractor.sync.barrier_*, its own Ractor wait/wakeup on a per-Ractor
condvar, no ready queue and no time... -
05:02 PM Bug #22263: NoMethodError: super: no superclass method 'require' for main — raised in a forked child process after a Ractor has exited in the parent (bisected to c59c4d717a)
- I've just encountered this issue with the following repro:
```
Ractor.new { 1 }.value
Kernel.module_eval { alias old_require require }
old_require 'tempfile'
```
Thanks @luke-gru for pointing this ticket out to me.
I ended u... -
04:30 PM Revision 46e53809 (git): [ruby/date] Implement compaction and embedded structs
- Currently dates are allocated in a 40B slot plus a 32B (simple)
or 48B (complex) external struct.
Once embedded they entirely fit into a 80B slot.
https://github.com/ruby/date/commit/e8a1442c60 - 04:00 PM Revision 89719849 (git): [ruby/date] Assert warnings for invalid offsets
- https://github.com/ruby/date/commit/569565d8c0
- 04:00 PM Revision c84cb4d4 (git): [ruby/date] Range-check the integral Rational offset
- Every branch of offset_to_sec range-checks the resulting number of
seconds except one path through the Rational branch: when the day
fraction is an integral Rational, n is assigned inside the if arm and
reaches *rof without passing the g... -
03:58 PM Revision 2084db2e (git): [ruby/date] Validate the commercial week range before conversion
- Reject week numbers outside -53..-1 and 1..53 before conversion so
integer overflow cannot make invalid input appear valid.
Fixes https://github.com/ruby/date/pull/180.
https://github.com/ruby/date/commit/e9c07010c7 -
- Pass the input encoding through `date__strptime` so non-ASCII trailing
data is treated as a parse failure instead of raising an encoding error.
Fixes https://github.com/ruby/date/pull/179.
https://github.com/ruby/date/commit/edf7cf016c -
- I found an unintended interaction with `Coverage.line_stub` after this change: calling `line_stub` while coverage is running can reset already-collected coverage counters.
I opened #22269 for this issue:
https://bugs.ruby-lang.org/is... -
- We keep a singleton of iseq wrappers (RubyVM::InstructionSequence). However,
this is rarely used and causes 8 bytes to be wasted on all iseq objects.
Since we now have the 32-byte slot size in the default GC, we can save that
memory. -
01:22 PM Bug #22278 (Open): `Coverage.line_stub` emits warnings found in source code
- `Coverage.line_stub` is implemented via `RubyVM::InstructionSequence.compile_file`. That emits all warnings found in the source code when `$VERBOSE = true`.
I found this via `simplecov`. It can do coverage for compiled erb, and it als... -
12:55 PM Bug #22276: alias in a module falls back to Object even in classes not inheriting from Object
- I implemented option 2: https://github.com/ruby/ruby/pull/18553
When `alias` in a module does not find the method in the module, it now creates a ZSUPER method entry whose original name is the aliased method. `Object` is searched only t... -
- When `alias` (or `alias_method`) is used in a module and the method is not found in the module, `alias` searches the method from `Object`. So a module can alias a method of `Kernel` or `Object`, and the alias works even if the module is ...
-
- ## Summary
[Github PR #18393](https://github.com/ruby/ruby/pull/18393)
`rb_gc_register_address` stores its entries in a single VM-wide list. This means that every Ractor local GC walks the list under a VM wide lock, which causes co... -
- https://bugs.ruby-lang.org/issues/21629
https://github.com/ruby/date/commit/214024627a -
- I agree with Jeremy, and would like to point out that Ruby already has exactly this semantics for refinements.
If the prepend in the example is replaced with a refinement, we get the expected behavior:
```ruby
using Module.new {
... -
05:03 AM Bug #22273: Aliasing doesn't interact well with Module#prepend
- As I mentioned in https://bugs.ruby-lang.org/issues/22263#note-5, I think a module should only be able to alias methods in ancestor modules, it should not be able to alias methods in descendant modules, as I don't think the semantics mak...
-
- This is a tracking issue for making RubyGems and Bundler work with Ruby::Box, so that reviewers can see the whole picture and what to look at next in one place.
## Goal
The goal is to realize [Feature #13847](https://bugs.ruby-lang... -
- Thank you for the report!
- 07:43 AM Revision 69b49ac7 (git): hash.c: avoid signed integer overflow in ar_hint_first_match SWAR splat
- This was previously flagged by ubsan because the AR_HINT_BASE_MASK
constant is implicitly typed as long:
hash.c:664:43: runtime error: signed integer overflow: 72340172838076673
* 131 cannot be represented in type 'long'
Notably, in an... -
-
-
- coroutine_thread_terminated() leaves the Ractor's living set before handing
the scheduler slot over, then keeps using th for the rest of the function:
thread_sched_to_dead_common() reads th->nt and deregisters th from the
barrier, and th... -
- MSYS2 dropped perl from its base install and the windows-2022 runner
image picked that up in 20260818.277.1, so autogen.sh now dies with
`/usr/bin/perl: bad interpreter`. master and ruby_4_0 set up their own
MSYS2 with autoconf, which pu... -
- MSYS2 dropped perl from its base install and the windows-2022 runner
image picked that up in 20260818.277.1, so autogen.sh now dies with
`/usr/bin/perl: bad interpreter`. master and ruby_4_0 set up their own
MSYS2 with autoconf, which pu... -
- Backport of the August 2026 resolv security release.
CVE-2026-80212
CVE-2026-80213
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> -
- test_global_variables called nonexistent assert_not_include?, raising
NoMethodError when the suite runs with RUBY_BOX=1.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> -
- Keep complete dependency rules in upstream gems while storing only source
mappings in the Ruby repository. -
- Keep generated rules even when manual rules cover them so source
mappings can round-trip between default gems and ruby/ruby. -
- The root cause is that `Marshal.load` is a builtin method written in Ruby (`marshal.rb`), so it is defined in the master box. The box resolution stops at that frame and looks up the dumped class names there, where user-defined classes ar...
- 12:36 AM Revision 888a9fe2 (git): [ruby/rubygems] Document :credential_store in the gemrc option list
- The option has been readable from gemrc since it was added, but the RDoc
list of valid gemrc options never gained a row for it.
https://github.com/ruby/rubygems/commit/3521da48ae -
- Alternation binds looser than the anchors, so `/\Atrue|false\Z/` reads as
`(\Atrue)|(false\Z)` and quietly turned every gemrc string starting with
`true` or ending with `false` into the boolean false. `:credential_store:
truestore` disab... -
- They sit after `private`, which does not apply to constants, so both are
public API today. Nothing outside the file reads them, and the setting
has not shipped, so this is the last release where making them private
costs nothing.
https:... -
- The two sides of one setting now disagree on `off`, so the list has to
say why it stays wider here. Nothing on the way in turns `off` into a
boolean, and nothing on this side shares the single vocabulary that
Bundler's boolean settings a... -
- `credential_store` was the only Bundler setting that read `off` as
disabled, while every other setting goes through `Settings#to_bool`,
whose false vocabulary has never had it. The disabled branch now
delegates to `#to_bool` so the two c...
08/27/2026
-
- A PLUGIN SOURCE section sends Bundler::Plugin.from_lock looking for the
plugin that handles it, and finding one loads and runs that plugin's
`plugins.rb`. The plugin index is read from the lockfile directory, so a
repository could run co... -
- `defined?(previous_root)` is true even when the assignment never ran, so
a failure before the root was saved reset Bundler's root to nil.
https://github.com/ruby/rubygems/commit/b2177cd67d
Co-Authored-By: Claude Fable 5 <noreply@anthro... -
- Bundler::Source::Rubygems stores remotes in reverse of the lockfile
order, so reading them straight through flipped which remote a spec
downloads from when a GEM section lists more than one.
https://github.com/ruby/rubygems/commit/da017... -
- The parser removed in #9564 turned a `!` entry in the DEPENDENCIES
section into a `= <version>` requirement. Bundler's LockfileParser
records the pinned source on the dependency and leaves the requirement
at `>= 0`, so `gem install -g` c... -
- Merge the unmerged dependency update at the previous merger.
https://github.com/ruby/io-console/commit/e712668ca2 -
- The save side matched the root as a string prefix, so an absolute path
that starts with the root but climbs above it was written back relative,
and the filter above then dropped it on the next load. Asking the same
question both ways lea... -
- A path stored relative to the plugin root only means anything inside it,
so an entry that climbs out of it did not come from a plugin Bundler
installed there. Paths recorded as absolute are left alone, which is how
`bundle plugin install... -
- +1 Kevin's got a lot of experience with TruffleRuby that can help improve CRuby
-
11:00 PM Misc #22272: Propose Kevin Menard (@nirvdrum) as a core committer
- +1.
> He's made significant contributions to ZJIT. Having Kevin as a committer would help the ZJIT development a lot.
Yes, it would be great if he can merge patches -
- +1 Kevin is a very experienced and thoughtful developer. He'd be a great asset to the CRuby committers.
-
04:34 PM Misc #22272: Propose Kevin Menard (@nirvdrum) as a core committer
- Thank you for the nomination, @tekknolagi, and the support from others. I'd be honored to accepted as a committer.
-
04:32 PM Misc #22272: Propose Kevin Menard (@nirvdrum) as a core committer
- +1 Kevin would be a great addition
-
04:18 PM Misc #22272: Propose Kevin Menard (@nirvdrum) as a core committer
- +1
-
04:05 PM Misc #22272: Propose Kevin Menard (@nirvdrum) as a core committer
- +1. He's made significant contributions to ZJIT. Having Kevin as a committer would help the ZJIT development a lot.
-
03:11 PM Misc #22272 (Open): Propose Kevin Menard (@nirvdrum) as a core committer
- ## Proposal
I'd like to propose Kevin Menard (@nirvdrum) as a committer. He started full-time work on ZJIT about a year ago and has been doing excellent work since. For example, he wrote the function inliner, which is the single bigge... -
- We cannot have YJIT and ZJIT enabled at the same time, so we can use a
single jit_payload pointer in iseq_constant_body to reduce the size by
8 bytes. -
-
09:17 PM Feature #22274 (Open): Make `IO::Buffer` no longer experimental.
- `IO::Buffer` was introduced in Ruby 3.1 by Feature #18020 as an experimental API. It provides an efficient buffer abstraction for fiber scheduler I/O, zero-copy operations, binary protocol implementations, and access from native extensio...
-
- If the two preceeding instructions are known not to be dependent
on the state of the stack, then we can eliminate the `swap` and inverse
the two previous instructions.
This mostly eliminate instruction in two cases:
```ruby
Object.new
... -
06:24 PM Bug #22273 (Open): Aliasing doesn't interact well with Module#prepend
- Currently, aliasing doesn't interact well with `Module#prepend` in my opinion.
### Example
```ruby
module Kernel
prepend(Module.new do
def require(feature)
puts "requiring feature (prepend): #{feature}"
super... -
- Here's an example of Rational#to_f giving an incorrect result when the numerator is a bignum and the denominator is a fixnum that can't be exactly represented as a double:
`Rational(10**24, 10**17 + 9).to_f` should return 10000000.0, but... -
- Consider the related code in `fix_fdiv_double`, which demonstrates a check for the denominator y being a fixnum that's exactly represented by a double.
```
static double
fix_fdiv_double(VALUE x, VALUE y)
{
if (FIXNUM_P(y)) {
... -
- > This bug is not related to Feature #21308 except for the same author.
Yes, I confused it with another ticket, that's why I deleted my comment, sorry. -
- This bug is not related to Feature #21308 except for the same author. The implementation of Rational#to_f does not perform any base conversions.
The bug is caused by the assumption that a floating-point conversion of a bignum numerator ... -
- Previously we were just checking the token, which meant something
like `{ a: 1, **{a: 2}.select { false } }` would warn. As esoteric
as this is, it was possible. This also happily gets rid of the
current_hash_keys field on parser, making... -
- Prevent file descriptor leaks when process creation exits through an
exception or throw.
https://github.com/ruby/open3/commit/8f5d4b6ae0 -
-
- The resolv security releases of 2026-08-27 fix CVE-2026-80212 and CVE-2026-80213. https://www.ruby-lang.org/en/news/2026/08/27/multiple-vulnerabilities-in-resolv/
`resolv` is a default gem. master already has the fixes through the aut... -
- Backport of the August 2026 resolv security release.
CVE-2026-80212
CVE-2026-80213
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> -
12:56 PM Bug #19865: Segfault when calling user signal handlers during VM shutdown
- I've just been analysing these recurring segfaults happening from time to time and landed on the bug report. The same combo stands: ruby-odbc and a connected db object at shutdown. Looking forward to a fix.
-
12:15 PM Bug #21830: Ruby::Box and Kernel#require
- Thanks @hsbt.
For the archives, I'd like to add something to
> This breaks Zeitwerk's implicit namespaces
The main idea in Zeitwerk is:
1. Scan the filesystem
2. Set autoloads accordingly
3. React to constants being autolod... -
- The direct `Kernel#require` decoration now works on master (after #21881), but the autoload path is still broken: an autoload fired under `RUBY_BOX=1` bypasses the decorated `require` entirely.
```ruby
module M
def require(path)
i... -
-
- On 64-bit systems, we used 4 bytes to determine if there is a collision
in the entry. This is a waste because it only needs 1 bit.
This commit separates each item in rb_id_table into three parts: items
(8 byte elements), keys (4 byte el... -
- When a C extension is required inside a Ruby::Box, the temporary copy is named by flattening the full source path into a single file name (`/` replaced with `+`). If the extension lives under a deep path, such as a long `BUNDLE_PATH`, th...
-
- In a user box created with gems disabled, prepending to a built-in module inserts it on the wrong side of the target:
```ruby
box = Ruby::Box.new
box.eval('module BoxDecor; end; Kernel.prepend(BoxDecor)')
box.eval('Object.ancestors... -
10:35 AM Feature #15381: Let double splat call `to_h` implicitly
- Hi,
I’m bringing this up again, hoping for inclusion in 4.1. If not, it would be great to know whether this is something that could happen in the future or whether it’s too much of a breaking change at this point.
Thanks! -
10:27 AM Revision 89d3b11e (git): [ruby/openssl] Look the SSLContext up from the SSL_CTX in the ALPN/NPN callbacks
- ossl_sslctx_mark uses rb_gc_mark_movable, so the SSLContext relocates. Its VALUE
is stored in four places: the SSL_CTX's ex_data, and the callback argument of the
NPN advertise, NPN select and ALPN select callbacks. ossl_sslctx_compact u... -
- By guaranteeing that hints can't match `RHASH_AR_CLEARED_HINT`
we can save on checking the key for `UNDEF_P` in a bunch of places. -
- Patch: https://github.com/ruby/ruby/pull/18533
-
- `Coverage.line_stub` compiles the file with `compile_file` only to read the ISeq's `trace_points`. Since https://bugs.ruby-lang.org/issues/22018 `compile_file` registers the file into the coverage table, so calling `line_stub` while `Cov...
-
- [Feature #21308] was filed a while ago.
`dtoa.c` is really showing its age, and in the last couple years there has been a lot of new algorithms on that particular topic.
Currently `ruby/json` uses Grisu for `double -> string` and E... -
- In iseq_scan_bits, we only update original_iseq if the element at code
moved. However, this isn't correct as it's possible that code has already
gotten updated but not original_iseq. This will cause original_iseq to not
get updated. -
- `Bundler.load_gemspec_uncached` returns nil when the gemspec evaluates
to nil, which an empty file does. `CLI::Init` was the only caller not
guarding against that, so it crashed with a `NoMethodError` backtrace
after already writing a pa... - 03:39 AM Revision 3d139905 (git): ZJIT: Eliminate constant-valued block params (#18369)
- In the following code,
```ruby
def test(cond)
if cond
x = 1
else
x = 1
end
x
end
```
the `x` after the conditional would ordinarily get a block param merging two constant `1`. Even if we had dominator-based GVN, this wo... -
- The `prune` vocabulary is being reused for a new setting, and `no_prune`
was already a poor name for a flag that only decides whether `bundle
cache` keeps gems that dropped out of the resolution. `no_prune` is
still read until Bundler 5 ... -
- Every documented spelling of the flag raises before reaching it, so the
only way in was Thor resolving `--skip-no-prune` to a negated value,
which `flag_passed?` does not match. That let an undocumented spelling
change a setting the docu... - 02:30 AM Revision 55379274 (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 2.86.6 to 2.86.7
- [Release notes](https://github.com/taiki-e/i... -
- We shouldn't record JIT information before the JIT is actually enabled.
This change doesn't seem to affect ZJIT or YJIT performance beyond the
margin of error:
ZJIT:
-------------- ------------ ------------ --------------... -
02:15 AM Revision 4dd33ca6 (git): ZJIT: Fix lattice bug in SSA minimization (#18482)
- Block param minimization was introduced here #17311, but a few tests did not properly eliminate redundant block params.
This is because I used `Vec::resize`. For most passes this was fine, especially when trivial block params were discov... -
02:01 AM Bug #22173: `Ractor.new { Fiber.current }.value` raises `[BUG] rb_ractor_confirm_belonging object-ractor id:1, current-ractor id:2`
- This assertion failure does not reproduce since https://github.com/ruby/ruby/commit/09aab7899fa0d0014e458c319e689175881a2075 so I think we can close it.
-
- Generic#== accepted only direct subclasses while Message compared any
descendant, so an instance of a subclass of a generated class compared
unequal. Both now go through Generic.type_class_equal?. Passing a block
to zip also avoids alloc... -
- Dropping the class cache made every decode build a fresh class for an
unknown type, so comparing the classes by identity reported two otherwise
identical resources, questions, and messages as different.
https://github.com/ruby/resolv/co... -
- https://github.com/ruby/resolv/commit/95e0a6664c
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> -
- Decoding a response with unknown (type, class) pairs or unknown
SvcParamKeys generated an anonymous class per value and registered it in
a constant and in ClassHash permanently, so a spoofed or malicious
response carrying many unknown va... -
- The comment predates the check moving into Label::Str and read as though a
label of 64 octets already overflowed its length octet. It does not: 64 to 255
write a reserved or compression pointer value, and only 256 or more wrap. Say
which... -
- Checking in Name.create left two ways to build an over-long label: Name.new,
which Config#generate_candidates uses to append search domains, and the
search list itself, which Config splits without going through Name.create.
Label::Str is... -
- A hostname is runtime data rather than a programming mistake, and callers
already wrap name resolution in rescue Resolv::ResolvError. Rejecting an
oversized name with ArgumentError escaped that rescue, since
Config#generate_candidates ru... -
- The counter started at -1, which measures the presentation form, so a name
encoding to 257 octets passed the 255 octet limit of RFC 1035 section 3.1.
The encoded form includes each label's length octet and the root label's
terminating ze... -
- A label longer than 255 octets wrapped its length octet mod 256 while the
data was written unchanged, so the wire bytes decoded to a different name
than the caller asked for. A hostname that passed an application's
allowlist could be sen... - 01:42 AM Revision efa87443 (git): Revert "[ruby/rubygems] Skip the RUBY_BOX gem CLI canary on Windows"
- This reverts commit 86817cf6ef9c1bcf7ef0c876ac1ead5b51c0b029.
- 01:42 AM Revision 1c664d80 (git): Defer unloading box-local extension DLLs on Windows
- box_entry_free could run before other T_DATA objects are finalized in
rb_objspace_call_finalizer, and FreeLibrary there unmapped the
rb_data_type_t and dfree functions that later finalization still
dereferences, crashing any boxed proces... -
-
-
- This reverts commit c0a526dd97b22f462f49a2a0062420523303769d.
-
08/26/2026
-
- 09:52 PM Revision 0968fd4b (git): Require `Fiber::Scheduler#fiber_interrupt`. (#18513)
-
- `eval_gemspec` computed `path.expand_path` inside its `chdir` block, so a
relative path resolved against the gemspec directory instead of the
original working directory, and loading `foo/foo.gemspec` handed `eval` a
file name of `<root>/... -
- A local override points a git source's install path at the developer's
own working repository, so scoping the removal to the definition's git
sources aimed `rm -rf` at their `.git`. Take only sources that live under
`Bundler.install_path... -
- Without one Bundler installs into the system gem directory, where the
cache is shared with RubyGems and holds gem files Bundler never put
there, so pruning there would delete other people's archives. Skip it
with a warning instead. Git m... -
- The forced-reinstall check walked the whole resolution, so a lockfile
carrying other platforms or an excluded group sent every `--redownload`
back to the remote. Scope it to what this platform and these groups
install, and let `bundle ca... -
- `bundle cache` installs first and copies the gem files afterwards, so
pruning at the end of that install left it with nothing to copy and it
failed with "Missing gem file". Skip the prune during that install and
run it once the app cache... -
- An installed gem resolves without a remote, so `bundle install
--redownload` had nothing to reinstall from once the cache was gone, and
`prune` left no way back. Go remote when a forced reinstall finds an
uncached gem, and look the gem u... -
- Container images ship Bundler artifacts they will never use again, so
images today hand-roll `rm -rf` over the cache and the git checkouts.
The new `prune` setting removes the same set from Bundler itself, and
everything it removes is re... -
09:27 PM Revision f3a72fe0 (git): [Bug #22242] Invalidate CCs on cc_table_dup
- When a Ractor is started, rather than modifying the cc table in place we
make a copy and CAS the table. When making the copy we previously
skipped cc_entries with an invalidated CME. However, that left the CCs
associated with that entry ... -
- Applied in changeset commit:git|f95e71cc1e2a082393d58db1eb6f57b09694f8cb.
----------
[Bug #22242] Invalidate CCs on cc_table_dup
When a Ractor is started, rather than modifying the cc table in place we
make a copy and CAS the table. Wh... -
- This reverts commit b746efc1fb46074ae659b4af0f9114aae9540e41.
-
- This reverts commit 9aa30b512f12ac3e298f7b971cdbd0e2cae8fc68.
-
- When a Ractor is started, rather than modifying the cc table in place we
make a copy and CAS the table. When making the copy we previously
skipped cc_entries with an invalidated CME. However, that left the CCs
associated with that entry ... -
- RDoc 8 parses rbs (including inline) and so
automatically removes the annotations for us
https://github.com/ruby/prism/commit/5f4d156205 -
-
- luke-gru (Luke Gruber) wrote in #note-4:
> [@jeremyevans0 (Jeremy Evans)](/users/1604) is this intentional or a bug?
This is general question of how aliased methods should work with prepend. Currently, we allow a class to alias a met... -
- I have a reproduction showing the issue:
```ruby
module M
def foo
print "M#foo(orig)"
end
end
class C; include M; end
M.prepend(Module.new do
def foo
print "P#foo("
super
print ")"
end
end)
C.new.... -
- Correction on the bisect result. 98f0921f2b is where the forked RailtieTest children start to crash (SIGSEGV or ArgumentError), not where this NoMethodError starts. Bisecting for the NoMethodError itself, with the same bin/test command, ...
-
- This commit adds support for using AR tables for identity hashes (hashes
created by method Hash#compare_by_identity). Before this change, identity
hashes always used ST tables. However, this used a lot more memory because
ST tables are q... -
11:56 AM Revision 68abb349 (git): [ruby/openssl] bn: support marshalling
- This fixes tests with Ruby master, where unshareable T_DATA is copied
across Ractor via Marshal.
https://github.com/ruby/openssl/commit/943fe2340e -
- https://github.com/ruby/mmtk/commit/3926721638
-
- Dropping Gem::Net::HTTPNotModified from it, or narrowing HTTPSuccess to
HTTPOK, left the suite green even though Updater depends on both the
304 and the 206 reaching it untouched.
https://github.com/ruby/rubygems/commit/b265de4bf7
Co-A... -
- Gem::RemoteFetcher#request lets SocketError, Errno::ECONNREFUSED and
their kin through unwrapped, and every caller here rescues FetchError
alone. On a host where only index.rubygems.org is unreachable, the
fallback to the Marshal index n... -
- Gem::RemoteFetcher#fetch_http has refused a redirect from https to plain
http since 2012, but the compact index path, which is now the default
and decides which versions resolve, followed it. One 302 from a
compromised edge or a misconfi... -
- Share the same code between gc.c and hash.c.
-
- On ruby-head, the Active Support (Rails) test suite fails with two Ractor-related errors (rails-nightly build 4651, https://buildkite.com/rails/rails-nightly/builds/4651#01a016ac-eada-492d-9548-44089079b973/L8324).
Rails CI failure. T... - 07:05 AM Revision 346b261a (git): Update bundled gems list as of 2026-08-26
-
- Ruby warns for duplicate keywords in the same literal hash:
```shellsession
$ ruby -we '{h: 1, h: 1}'
-e:1: warning: key :h is duplicated and overwritten on line 1
```
Ruby does not warn for duplicate keywords in splatted hashes... - 06:04 AM Revision 4a6678f6 (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 2.86.5 to 2.86.6
- [Release notes](https://github.com/taiki-e/i... -
- A boxed gem CLI subprocess dies with SIGSEGV in
rb_objspace_call_finalizer at exit on Windows. The fix belongs in
Ruby::Box itself, so omit the canary there until it lands.
https://github.com/ruby/ruby/actions/runs/32933358242/job/98069... -
- Applied in changeset commit:git|e5132b162737829f744b2b23908b9d08479db977.
----------
[Bug #22264] Warn for duplicate keys after leading keyword splat
Leading keyword splats must not suppress duplicate literal key
warnings. -
- I agree with Jeremy's concern, but I think it should be discussed separately.
As for this bug, it stems from an oversight when we allowed `**` to be placed in locations other than the end of a hash literal; it should be fixed simply by ... -
- Leading keyword splats must not suppress duplicate literal key
warnings. -
- The unit tests cover each fix on its own, but nothing checks that a boxed
gem command actually gets as far as listing gems.
https://github.com/ruby/rubygems/commit/e6ee031539
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> -
- Under Ruby::Box mkmf makes RbConfig.expand recurse until SystemStackError,
so building a native extension failed even once the CLI itself started.
Build tools are not a target of box isolation.
https://github.com/ruby/rubygems/commit/b4... -
- Under Ruby::Box (RUBY_BOX=1) the builtin Marshal runs in the root box and
cannot resolve Gem:: constants defined in the main box, so every gem
command failed to start while deep-copying its option defaults.
https://bugs.ruby-lang.org/is... -
- When several gems come from the same git block, Runtime#cache invokes
the shared Source::Git once per gem, and cache_to re-cloned and
re-checked out the whole worktree every time. For large repositories
this made bundle cache/update/inst... -
- TOPLEVEL_BINDING always belongs to the main box, so when Bundler is
loaded inside a user box, gemspecs were evaluated against the main
box's Gem::Specification, which lacks Bundler's monkey patches and
fails with NoMethodError on `source... -
- Content-Length parsing became strict in 4.1, so values with anything
around the digits now raise instead of returning the digits.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> - 12:46 AM Revision 2c260b8d (git): [ruby/net-http] Make Content-Length handling strict
- (https://github.com/ruby/net-http/pull/324)
`Content-Length` value must have only digits.
See also RFC 9110 Section 8.6:
https://www.rfc-editor.org/rfc/rfc9110.html#section-8.6
Content-Length = 1*DIGIT
The current implementation ... -
- A peer that keeps sending 1xx responses keeps the client in the read
loop of transport_request indefinitely, since the per-response header
limit resets for each response. Cap the count at 100, the same value
CPython's http.client uses.
... -
- Ruby 3.3 and 3.4 can allow an untimed `Thread#join` to return after a spurious Fiber scheduler wakeup, even though the target thread is still alive.
Because `Thread#value` relies on `Thread#join`, it can consequently return `nil` befo... -
- symbol_id_entry_directory is not an embedded TypedData, so we need to free
the allocated id_entry_dir as well.
08/25/2026
-
- The response header cap added by #219 applies only once a complete line
has arrived, so a server that never sends a line terminator can still
grow the read buffer without bound, as reported in #315. Read the
status line, header lines, ch... - 10:50 PM Revision b5b032cc (git): Update default gems list at 7f00b21798a9dc156f43ab6fed9505 [ci skip]
-
- https://github.com/ruby/net-protocol/commit/fb347d5ff4
-
- I'm not sure whether it is desirable to warn for `{a: 1, **{a: 2}}`. I would guess that internally, this hits the same warning for `{a: 1, a: 2}` (which should warn), but for implementation details, it doesn't hit the warning for `{**{},...
-
- $ ruby --parser=parse.y -cve '{**{}, a: 1, **{a: 2}}'
ruby 4.1.0dev (2026-04-21T02:07:14Z aot be86d5704f) [arm64-darwin23]
Syntax OK
$ ruby --parser=parse.y -cve '{a: 1, **{a: 2}}'
ruby 4.1.0dev (2026-04-21T02:07:14Z aot be86d5704f)... -
09:55 PM Revision 9270124f (git): [ruby/rubygems] Delete `Bundler::ParallelInstaller::SpecInstallation#enqueued?`
- - This predicate method has been unused since
https://github.com/ruby/rubygems/commit/7d1131672a84 in 2021 when the
`ready_to_enqueue?` method was simplified.
https://github.com/ruby/rubygems/commit/88d7eef941 -
- - In https://github.com/ruby/rubygems/commit/304a98301522 in 2013, "quiet?" was any
level that was not "debug", "info", or "warn",
- Its last use was removed in https://github.com/ruby/rubygems/commit/64f1294672e2
in 2014.
- Since ... -
09:51 PM Bug #22261: ruby-head builds have been failing on aarch64-linux CI in ruby/ruby-dev-builder
- The daily "CRuby Dev Builds" workflow in ruby/ruby-dev-builder has been successful. Thanks for your quick fix!
-
- Applied in changeset commit:git|e77a77111224426ca18eaeff55d7ad1618cbacc9.
----------
[Bug #22261] Keep GC root-marking frame alive
Prevent a tail call from discarding saved machine-register roots
before `rb_gc_mark_roots` scans the mac... -
- nobu (Nobuyoshi Nakada) wrote in #note-1:
> Does [GH-18495](https://github.com/ruby/ruby/pull/18495) fix it?
Thank you for creating a fix.
The easiest way to know would be to merge that PR, and `ruby-dev-builder` runs once per day.
... -
- Does [GH-18495](https://github.com/ruby/ruby/pull/18495) fix it?
-
- The daily "CRuby Dev Builds" workflow in ruby/ruby-dev-builder has failed every day since 2026-08-07:
* https://github.com/ruby/ruby-dev-builder/actions/workflows/build.yml
Because of this, `ruby-version: head` currently installs... -
- Where a credential ends up, what happens when the store cannot be reached,
that the store belongs to the machine rather than to a project, and that
credentials already written in plain text are not migrated on their own.
https://github.... -
- Bundler reaches Gem::CredentialStore by name, so a bundler installed
against an older RubyGems needs the file itself, the way it already
carries the YAML serializer and the compact index client. Three places
enumerate what bundler ships,... -
- The credential_store setting takes the same values on this side, and
credential_store.<host> picks a different backend for one host, since a
backend gem published for one registry cannot know what its users want
everywhere else. Bundler ... -
- Signing out has to reach both places a key can live, including the keys
saved for other hosts with gem signin --host. The two halves report
separately, so a credentials file that cannot be removed does not leave
the stored keys behind, a... -
- The :credential_store: gemrc setting, off by default, moves API keys out
of ~/.gem/credentials and into the OS store. A key is read from the store
ahead of the file, so the two never disagree about which one wins, and the
plain text copy... -
- Storing an API key in a plain text file leaves it readable by anything
running as the user. This wrapper keeps secrets in the operating system's
own store instead, through the security CLI on macOS, secret-tool on
Linux, and Windows Powe... -
- In `install`, `self.to_s` is called by `print_using_message "Using ... from #{self}`.
This in turn calls `current_branch` which if the local git cache
is missing will re-clone it from scratch.
This is an issue, because a common optimiza... - 07:38 PM Revision 095a2917 (git): ZJIT: Add codegen test for Fixnum#<< overflow side exit (#18390)
- Add codegen tests for the Fixnum left shift fastpath and its overflow side exit.
- 05:45 PM Revision 383397e1 (git): ZJIT: Index into instruction vectors with InsnId directly
- 05:45 PM Revision abc5a924 (git): ZJIT: Index into block vectors with BlockId directly
-
-
- You should open an issue on the google oss-fuzz repo, I didn't even know that thing existed. It is clearly wrong.
-
02:46 PM Misc #22262 (Third Party's Issue): OSS-Fuzz `fuzz_prism` target never calls `ruby_init()`, so Prism is not being fuzzed
- The OSS-Fuzz fuzz target `fuzz_prism` crashes on every non-empty input before it reaches
the parser, so Prism has never actually been fuzzed. The harness source lives in OSS-Fuzz
(`projects/ruby/fuzz_prism.cpp`), not in this repository... -
- `hash_copy` will call `st_init_existing_table_with_size` which
allocates and may trigger GC causing the current hash to be marked.
So we need to zero-out the `st_table` to avoid a segfault. -
- ### 1. First occurrence in Rails CI
https://buildkite.com/rails/rails-nightly/builds/4651#01a016ac-eaf9-4a71-9fb6-1924c8a7b9af
(job "activemodel (master-debug)", 2026-08-18, ruby 4.1.0dev (2026-08-18T13:12:35Z master 6b719acc57) +PRI... -
- Keep the payload builder uninlined in optimized GCC builds because
inlining causes false-positive `-Wclobbered` warnings. -
-
02:03 PM Revision 17026737 (git): Replace Array#flatten{,!} internal ident hash with set
- This used a hash with all true values, with is simple to convert
to a set to save on memory when flattening array with a large
number of nested elements. -
- This is designed to replace rb_ident_hash_new calls when such hashes
are used purely with with true values.
To implement this, add a set_alloc_with_size_and_type static function,
and have set_alloc_with_size and rb_ident_set_new call it. -
-
-
- Instead of always allocating `ar_table` backed hashes in 160B
slots for them to have capacity of `RHASH_AR_TABLE_MAX_SIZE` (`8`),
they can now be allocated in any slot size between `64B` and `160B`.
`64B` is the minimum slot size that s... -
- For now it still returns a constant, but clarify when we're looking
for the max bound, versus a miss, etc. -
- Prevent a tail call from discarding saved machine-register roots
before `rb_gc_mark_roots` scans the machine stack. -
- With gcc 15.2.0 on aarch64 Ubuntu 26.4:
```console
ractor_sync.c: In function ‘ractor_basket_new.constprop’:
ractor_sync.c:1093:82: warning: argument ‘type’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
1093 | ractor_basket_... -
- With gcc 15.2.0 on aarch64 Ubuntu 26.4:
```console
io_buffer.c: In function ‘io_buffer_flags_for_new’:
io_buffer.c:978:20: warning: comparison between ‘enum rb_io_buffer_flags’ and ‘enum <anonymous>’ [-Wenum-compare]
978 | if (all... -
- With gcc 15.2.0 on aarch64 Ubuntu 26.4:
```console
hash.c: In function ‘rb_hash_s_create’:
hash.c:1982:20: warning: ‘hash’ may be used uninitialized [-Wmaybe-uninitialized]
1982 | return hash;
| ^~~... -
- https://github.com/ruby/prism/commit/9df2a21995764880c6d27213e2c7960f2232c879 changed the comment style
Now rdoc picks this up. The comment doesn't say much, seems fine to just remove
https://github.com/ruby/prism/commit/f703c615ae -
- The floor's two reasons fit in one block, and the two tests that
restated them are named for what they cover.
https://github.com/ruby/net-protocol/commit/6ffd258b7c
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> -
- readuntil buffers until the terminator arrives, so a peer that never
sends one grows the read buffer without bound. The limit lets a protocol
implementation cap what a single read may return, raising the new
Net::ReadLimitExceeded instea... -
- So I just ran into that issue again on my PR to make `ar_table` dynamic: https://github.com/ruby/ruby/pull/18345
`Hash.new(capacity:)` now can't allocate anything but the smallest usable slot. In theory `Hash.new(capacity: 8)` should ... - 09:50 AM Revision a06bf2f8 (git): Lock `IO::Buffer` coercion callbacks. (#18490)
-
- rbs 4.2.0 adds Etc and IO#pathconf stdlib tests that need getpwent(),
getgrent() and the pathconf constants, none of which exist on Windows,
so they go in the Windows-only list. OpenURISingletonTest reaches
www.ruby-lang.org and now fail... -
- The bundled_gems workflow has failed every day since 2026-08-16 because
rbs 4.0.3's Ripper_test.rb no longer finds Ripper on master. rbs 4.2.0
picks up the upstream fix that requires "ripper" in the stdlib tests.
typeprof 0.32.0 never fi... -
- rbs was pinned in March 2026 while the 4.0 series was breaking CI. Now
that 4.2.0 is in gems/bundled_gems and the tests pass again, the daily
workflow can pick up later rbs releases on its own. -
- Cover a terminator split across two reads and across more than two, the
negative rewind the floor clamps, the floor keeping the search out of
consumed bytes, and the ignore_eof branch that returns data without a
terminator. FakeReadParti... -
- BufferedIO#readuntil resumed searching at the previous buffer end, so a
multi-byte terminator split across two fills, such as "\r" ending one
chunk and "\n" starting the next, was never matched and readuntil
returned data past the termin... -
- This bug seem to have been present all the way back since Ruby 2.5, so requesting a backport accordingly.
-
03:10 AM Bug #22259 (Closed): Arrays sharding a buffer segfault when concatenating with each other
- Applied in changeset commit:git|46c287e2ca39da63d5164e2b7227308a41554c9a.
----------
Fix splicing an array that shares its buffer with a longer array
Arrays sharing one buffer have the same data pointer, so rb_ary_splice()
took `a` in ... - 07:50 AM Revision 753ace46 (git): Lock `IO::Buffer` allocations during copy. (#18489)
-
- I've managed to reproduce this under rr locally through the Active Record test suite with `VM_CHECK_MODE` enabled. I don't have a more reliable reproduction yet or a fix.
What the recording seems to show is that the ccs entries become o... -
- The crash does not reproduce on master, so bisecting from the 4.0 branch point (d375bcc965d4) gives the first fixed commit:
> 1da540aeaa1507af4c050bf8a6ee2802e65f9aac "Invalidate CC on cme free as well" (gc.c, +3/-1)
The boundary i... -
06:20 AM Revision 45088e23 (git): [ruby/mmtk] Prevent overflow values
- (https://github.com/ruby/mmtk/pull/107)
https://github.com/ruby/mmtk/commit/f6b8b0d5ae - 04:55 AM Revision b9bcae99 (git): Preserve `IO::Buffer` read-only state when resizing. (#18487)
-
- Most callers have the replacement as an array object; let them pass it
directly.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> -
- Arrays sharing one buffer have the same data pointer, so rb_ary_splice()
took `a` in `b = a.dup; b.pop; b.concat(a)` for `b` itself and rebased
the source onto `b`'s shorter storage, copying uninitialized memory.
Have the callers tell wh... -
- The temporary file was renamed into place before being closed, so
buffered data had not reached the filesystem yet and a concurrent reader
could observe an empty destination. A flush failure such as ENOSPC would
surface only after the de... -
- The basename was cut with byteslice, which can split a multibyte
character in half. Filesystems enforcing UTF-8 validity such as APFS then
reject the temporary name with EILSEQ.
https://github.com/ruby/rubygems/commit/a7dc5647b0
Co-Aut... - 02:46 AM Revision 44da9d73 (git): Make `IO::Buffer` I/O operations single-transfer and composable. (#18483)
-
- Fixes the Style/RegexpLiteral offense from CI.
https://github.com/ruby/rubygems/commit/c4069df44d
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> -
- The fixture changelog still used the old `## VERSION (DATE)` header, so
the fixed `release_date_for` no longer matched it and the released case
of the version spec failed.
https://github.com/ruby/rubygems/commit/36ee802a7e
Co-Authored-... -
02:34 AM Revision 0a316984 (git): [ruby/rubygems] Fix release date extraction for the current changelog header format
- release_date_for in spec/support/build_metadata.rb matched the
changelog's old `## VERSION (DATE)` header, but .changelog.yml has
generated `## VERSION / DATE` headers since Bundler 4.0.9. The regex
silently failed to match, so built_at ... -
- This can be addressed by Ruby::Box [Feature #21311], which loads code, including C extensions, into an isolated space with its own constants and loaded features. That resolves the concerns raised in 2017 about shared library conflicts an...
-
-
12:34 AM Revision 3d36072a (git): [DOC] Fix `IO.puts` doc
08/24/2026
-
11:37 PM Bug #20243: M:N threading VM_ASSERT failure in rb_current_execution_context with clang 17 (on Linux)
- wanabe (_ wanabe) wrote in #note-3:
> 7866e124a852c344b5762eb917c03a1f95d9058d == https://github.com/ruby/ruby/pull/13217 fixed the issue.
I don't think it is fixed, please see https://gcc.gnu.org/PR124995. It happens with recent GCC... -
- 7866e124a852c344b5762eb917c03a1f95d9058d == https://github.com/ruby/ruby/pull/13217 fixed the issue.
```
$ git checkout 7866e124a852c344b5762eb917c03a1f95d9058d~ && make -j miniruby 2>&1 >/dev/null && ./miniruby ../a.rb 2>&1 |head -n... -
- Giving up part way through a frame leaves the stream between frame
boundaries, so reusing the cached requester made every later attempt
read the previous frame's bytes as a length prefix. Asking the requester
instead of its class keeps a... -
- Requester#request computes a monotonic time limit but only the socket
readability wait honoured it, so a peer that sends an incomplete frame
and keeps the connection open blocked past the configured timeouts
until it closed. The paramete... -
- Thx for the patch, much cleaner than mine. Indeed I thought there must have been a cleaner fix, but I admit I was out of my depth.
-
- Confirmed and reproduced. Thanks for the report.
While looking into it I found that the same line causes another bug.
```ruby
a = [:A, :B, :C, :D, :E]
b = a[1, 4] #=> [:B, :C, :D, :E]
a[0, 0] = b
p a # expected: [:B, :C... -
- A succession of crashes was observed on a particular workload from our app, involving recursive set operations, after upgrading it to ruby 4.0.6 . LLM was used to build a reproduction (shared in the description of the PR linked below), a...
-
- With the changelog dump gone, the pre-update notice is where users decide whether to update, so link the versioned CHANGELOG.md there too, matching npm's three-line notice.
https://github.com/ruby/rubygems/commit/4ca3c9eba1
Co-Authored... -
- CHANGELOG.md is over 7,000 lines of generated PR titles, and printing every entry since the previous version dominated the output (478 of the 499 lines on the ruby:3.4 image). Print a single line linking the versioned CHANGELOG.md instea...
-
-
- 08:50 PM Revision 177ead8d (git): Make IO::Buffer MemoryView ready (#17965)
- `null?` buffer isn't MemoryView available.
Not `null?` buffer is MemoryView available. It can be exported as a
1-dimensional contiguous array of bytes.
Users can't change the underlying memory of the exporting IO::Buffer
while exported... - 06:44 PM Revision c4fdf07d (git): ZJIT: Reduce SideExit HashMap rehash (#18411)
- If we don't pre-size the HashMap, it has to re-hash when resizing.
Hashing a SideExit is slow. Give a size hint so we only hash ~once per
exit. - 06:25 PM Revision 6e3061c7 (git): Add accessors for ISEQ_BODY(iseq)->variable fields
- 06:25 PM Revision 5552e2c5 (git): Shrink iseq->body struct: allocate iseq->body->variable lazily
- Most of the time in a production app most iseqs won't have iseq->body->variable
set to anything non-default. We can allocate it lazily and save 32 bytes per iseq.
Some large apps have millions of iseqs, so it adds up.
32B saved per iseq... - 06:13 PM Revision 90529c25 (git): Raise YJIT’s `case` specialization limit to 256 (#18374)
- Raise YJIT case dispatch specialization limit to 256
-
05:38 PM Feature #22231: Add `IO::Buffer#index`
- Thanks for taking this.
Matz asked for out-of-range `offset` and `length` to return `nil` rather than raise. Assuming you have no objection to that direction, there are a few points I would like to think through.
The first is wheth... -
- Thanks for creating this issue, I will review the details.
- 03:04 PM Revision 065a5592 (git): [ruby/json] Add JSON::ParserError#json_path
- (https://github.com/ruby/json/pull/1062)
Add JSON::ParserError#json_path
https://github.com/ruby/json/commit/9ae76bdd3d
Co-authored-by: Jean Boussier <byroot@ruby-lang.org> -
- We only use either positions and succ_index_table (when VM_INSN_INFO_TABLE_IMPL == 2,
positions is converted to succ_index_table), so we waste 8 bytes in
rb_iseq_constant_body. This changes it so that they are in a union. -
12:51 PM
Feature #21700 (Closed): `IO::Buffer.map`: offset argument is "broken" and needs to be made more useful
- Applied in changeset commit:git|53bc08564542206d0f93dfe984059e245ad16d46.
----------
Expose `IO::Buffer` mapping alignment. (#18474)
[Feature #21700] - 12:50 PM Revision 53bc0856 (git): Expose `IO::Buffer` mapping alignment. (#18474)
- [Feature #21700]
-
- Saves on potentially having to reallocate multiple times.
-
12:08 PM Bug #22260 (Open): Ruby::Box: SIGSEGV when calling a method inside an isolated Proc defined at class/module body (RUBY_BOX=1)
- ## Summary
Under `RUBY_BOX=1`, any method call inside a Proc isolated by `Ractor.make_shareable` (or `Proc#isolate`) and defined at a class/module body crashes the VM:
```console
$ RUBY_BOX=1 ruby -e 'module M; L = Ractor.make_sha... -
11:50 AM
Bug #22219 (Closed): IO::Buffer#free, #resize and #transfer should raise FrozenError if the receiver is frozen
- Applied in changeset commit:git|5ebbf33a289a314d424fd2b51f5476b0959f4ad3.
----------
Respect frozen `IO::Buffer` lifecycle state. (#18473)
[Bug #22219] - 11:50 AM Revision 5ebbf33a (git): Respect frozen `IO::Buffer` lifecycle state. (#18473)
- [Bug #22219]
-
11:32 AM
Bug #21672 (Closed): `IO::Buffer.new` does not check that flags are valid
- Applied in changeset commit:git|d3bf4529e3553f25104a2f80a5335e536e36bd90.
----------
Validate `IO::Buffer` constructor flags. (#18470)
[Bug #21672] - 11:32 AM Revision d3bf4529 (git): Validate `IO::Buffer` constructor flags. (#18470)
- [Bug #21672]
-
- Start slower and less predictable shards earlier to reduce the overall
workflow completion time. -
11:12 AM
Bug #22195 (Closed): IO::Buffer read after free silently returns empty data instead of raising
- Applied in changeset commit:git|281c2d4c19f3e7a10772ae0e62ba01ae3695500a.
----------
Document `IO::Buffer#free` empty state. (#18471)
[Bug #22195] - 11:12 AM Revision 281c2d4c (git): Document `IO::Buffer#free` empty state. (#18471)
- [Bug #22195]
- 11:00 AM Revision cc706f39 (git): Clarify `IO::Buffer` slice validity documentation. (#18469)
-
- It has been fixed at 1c97abaabae6844c861705fd07f532292dcffa74 == https://bugs.ruby-lang.org/issues/19907.
I confirmed it with arm32v7/ubuntu:noble docker image on qemu-user-static.
```
root@1294e0e49506:/ruby/tmp/master# for commit ... -
- A problem I ran into in production, is that the boot sequence,
which tend to be allocation heavy, would cause the `allocatable_bytes`
(or previous version equivalent) to grow much higher than what the
application runtime actually needs.
... -
09:24 AM Bug #22242: SEGV in method dispatch (`vm_call_iseq_setup_kwparm_nokwarg` / `def_iseq_ptr`) on Ruby 4.0.6 — Rails CI
- A `VM_CHECK_MODE` build turns this from a rare SEGV in random places into an
assertion at the moment the corrupted call cache is used, and the assertion
says the cc is holding a reclaimed cme.
```
RUBY_CONFIGURE_OPTS="cppflags=-DVM... - 09:06 AM Revision 09fdc3be (git): Avoid unsafe zero-length `IO::Buffer` operations. (#18465)
- - All `null?` buffers will result in zero-length operations.
-
- They flip the same strict flag as --filter-strict, so the cooldown
caveat applies to them too. Review finding UX-1 (round 3).
https://github.com/ruby/rubygems/commit/92933f7b7b
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> -
- Review finding UX-3 (round 2).
https://github.com/ruby/rubygems/commit/0fe5f1591e
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> -
- The resolver already pins its cooldown clock via cooldown_now; do the
same here so every annotation in one run shares the same reference
time. Review finding API-2 (round 2).
https://github.com/ruby/rubygems/commit/8a945d0d61
Co-Author... -
- Unreachable today because specs_for_outdated_check yields one spec per
name, but keying on the argument that feeds the computation keeps the
cache honest. Review finding API-1 (round 2).
https://github.com/ruby/rubygems/commit/e710e9ef9... -
- No test exercised the "1 more day" path. Review finding GEM-2.
https://github.com/ruby/rubygems/commit/a1baec1d4b
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> -
- In strict mode the resolved version is already cooldown-filtered, so no
cooldown annotations appear. Pin that down. Review finding UX-4.
https://github.com/ruby/rubygems/commit/6f564d219a
Co-Authored-By: Claude Fable 5 <noreply@anthrop... -
- Review finding GEM-1.
https://github.com/ruby/rubygems/commit/595c21856d
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> -
- The candidate list was recomputed for the same gem by
retrieve_active_spec and newest_out_of_cooldown. Review finding API-2.
https://github.com/ruby/rubygems/commit/1be428df41
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> -
- The man page enumerates the cooldown output tokens, so the two tokens
added for #9624 belong there too. Review finding UX-1.
https://github.com/ruby/rubygems/commit/9d227a2b10
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> -
- The table said "1.5.0 available", but "available" already denotes the
opposite state in the install summary, which reports a blocked version
as "(available in 6 days)". Reuse the exact antonym of the prose
"in cooldown for N more days" i... -
- With cooldown enabled, "newest" is often a version still inside the
cooldown window and the newest version bundler would actually adopt was
not shown anywhere. Append it to the outdated line, both in the
parseable output and in the table... -
- Bundler runs against whatever RubyGems the host provides, so calling `Gem::Util.glob_files_in_dir` left `bundle cache` pruning able to resolve a cache entry named `~` to the real home directory and hand it to `FileUtils.rm_rf` on every R...
-
- `gem contents`, `gem stale` and the `gem setup` old-doc cleanup returned no matches when the install path contains glob metacharacters. For `--lib-only`, glob `raw_require_paths`, which stay relative to `full_gem_path`, rather than `requ...
-
- Matched entries were resolved with File.expand_path, so an entry starting with `~` was expanded into the home directory. A cache directory named `~` made `bundle cache` pruning delete files under `$HOME`, and an entry like `~foo` raised ...
-
- A created_at without a time zone offset was parsed as local time, so a
third-party server that omits the offset shifted the cooldown window by
whatever offset the machine running bundler happened to have.
rubygems.org always sends one, s... -
- A cooldown from a config file or from BUNDLE_COOLDOWN went through
value.to_i, so "abc" became 0 and "-5" stayed negative, and either one
disabled the cooldown without saying so. Read it with Integer() and
warn, naming the effect: such a... -
- The resolver asks for the cooldown once per candidate spec, paying a
Bundler.settings lookup each time, so read it once per remote. That
makes the value a snapshot, which is fine because nothing changes the
setting after the sources are ... -
- The gmp portfile at the previous baseline downloads
autoconf2.71-2.71-3-any.pkg.tar.zst from msys2 mirrors by a pinned URL,
but msys2 has rotated the package to -4, so every mirror returns 404 and
Windows CI fails to build gmp whenever t... -
- Rather than start from a mutable hash created with `newhash` or
`duphash`, and then iteratively insert into it, it's preferable to
start from a constant hidden hash, so that we can directly allocate
the final hash with the right size.
F... -
- Keep the source artifact when compilation fails so failed shards can be
re-run. Allow source preparation itself to overwrite the artifact. - 04:46 AM Revision 4c978a5b (git): Make empty `IO::Buffer` slices valid. (#18461)
-
- Avoid separate builds for checks already enabled by `RUBY_DEBUG` or
`RUBY_DEVEL`. - 04:04 AM Revision 4d5ae426 (git): Document dynamic `IO::Buffer` slice validity. (#18459)
- 03:04 AM Revision 98606e21 (git): [ruby/rubygems] Document Bundler proxy semantics in bundle-config
- The switch from net-http-persistent to Gem::Request (https://github.com/ruby/rubygems/pull/9786) aligned
Bundler proxy handling with RubyGems: https_proxy now takes precedence
for https sources, proxy credential variables follow the sche... - 03:04 AM Revision cc4a4828 (git): [ruby/rubygems] Restore NO_PROXY wildcard support in Bundler proxy detection
- net-http-persistent treated NO_PROXY="*" as a request to bypass the
proxy for every host, but Gem::URI::Generic.use_proxy? parses "*" as a
literal hostname that never matches, so the wildcard stopped disabling
proxies after the switch to... - 03:04 AM Revision e6e55f97 (git): [ruby/rubygems] Disable transport-level retries on Bundler connections
- Gem::Request already resends requests on stale connections and
Bundler::Retry retries failed requests, so the Gem::Net::HTTP retry
layer (max_retries defaults to 1) only multiplied the number of
attempts, doubling the worst-case wait on ... - 02:47 AM Revision 10465356 (git): [ruby/rubygems] Preserve native extensions during implicit clean
- https://github.com/ruby/rubygems/commit/3abbef3c18
- 02:34 AM Revision f9f83168 (git): Bump the github-actions group across 1 directory with 4 updates
- Bumps the github-actions group with 4 updates in the / directory: [github/codeql-action/init](https://github.com/github/codeql-action), [github/codeql-action/analyze](https://github.com/github/codeql-action), [github/codeql-action/upload...
- 02:19 AM Revision e2044876 (git): [ruby/rubygems] Honor NO_PROXY="*" in Gem::Request::ConnectionPools
- curl and net-http-persistent treat a "*" entry as bypassing the proxy
for every host, and Bundler's fetcher already honors it, but no_proxy?
never matched it, so gem commands kept using the proxy.
https://github.com/ruby/rubygems/commit... -
02:14 AM Bug #22258 (Open): Update bundled json to >= 2.19.2 on ruby_4_0 (CVE-2026-33210)
- The ruby_4_0 branch bundles json 2.18.0 (ext/json/lib/json/version.rb), which is affected by CVE-2026-33210 / GHSA-3m6g-2423-7cp3 (format string injection when parsing with allow_duplicate_key: false, fixed in json 2.19.2 — there is no p...
-
- Read the previous method entry before slow lookup invalidates its call
cache. This allows debug counters and assertions to be enabled
together. -
-
-
-
-
-
-
- - `Gem::Security::Signer#extract_name` matched `/\Aemail:/` then read
the non-matching part of the string out of `$'`.
- Prefer `delete_prefix("email:")` as it's understandable by mortals.
- Consequently get rid of a RuboCop disable ... -
- We know the type of the proc object at allocation time, so we can allocate
exactly the amount of memory needed for that type.
We can also shrink the type from a 4 byte enum since there are only 4 types
of procs. This allows us to put so...