Activity
From 04/02/2026 to 04/08/2026
Today
-
02:34 PM Feature #21982: Add `Decimal` as a core numeric class
- Oh, gotcha. Yeah we would not be able to optimize C code. Maybe one day it will motivate a pure Ruby impl but today is not that day.
-
07:06 AM Feature #21982: Add `Decimal` as a core numeric class
- > ZJIT will be able to "see through" and even elide general Ruby object allocations when possible.
I assume this will remain limited to objects and methods defined in Ruby? `BigDecimal` is implemented in C, and I don't think it could... -
02:17 PM Bug #21984: RubyVM::AST can't tell if an array is passed to break/next/return
- I don't think that fix works:
```
$ ./ruby -e 'pp RubyVM::AbstractSyntaxTree.parse("tap { break 1, 2 }")'
(SCOPE@1:0-1:18 tbl: [] args: nil body: (ITER@1:0-1:18 (FCALL@1:0-1:3 :tap nil) (SCOPE@1:4-1:18 tbl: [] args: nil body: (BREAK... -
04:53 AM Bug #21984: RubyVM::AST can't tell if an array is passed to break/next/return
- Changing this causes the rbs tests to fail.
https://github.com/ruby/ruby/actions/runs/24117591586/job/70364654030?pr=16678#step:14:828
``` diff
class Hello
- def with_return: () -> (1 | "2" | :x)
? ^
+ ... -
- ```
irb(main):005> RubyVM::AbstractSyntaxTree.parse("tap { break [1, 2, 3] }")
=> (SCOPE@1:0-1:23 tbl: [] args: nil body: (ITER@1:0-1:23 (FCALL@1:0-1:3 :tap nil) (SCOPE@1:4-1:23 tbl: [] args: nil body: (BREAK@1:6-1:21 (LIST@1:12-1:21 (... -
01:00 PM Revision 7209523f (git): [ruby/openssl] kdf: fix wrong OPENSSL_cleanse() calls
- Embarrassingly, the previous commits introduced OPENSSL_cleanse() calls
against the temporary struct instead of the buffer content. Thanks to
nagachika for noticing.
https://github.com/ruby/openssl/commit/8eca3efad4 -
- scrypt is another password hashing algorithm, so releasing the GVL is
useful.
https://github.com/ruby/openssl/commit/dd2f6ba892 -
- Since PBKDF2 runs single-threaded and is typically configured to take
several hundred milliseconds or longer, it is a perfect candidate to be
run without the GVL.
https://github.com/ruby/openssl/commit/2a24966414 -
- `imemo_tmpbuf` can be skipped if `ptr` is `NULL` which is often
the case if `RB_ALLOCV_END` was used. -
-
- Applied in changeset commit:git|24a2ba09af13b7c969733ea9370ad59d2442f4c9.
----------
[Bug #21985] Include the `-` in the negative numbers location -
- ```
irb(main):007> RubyVM::AbstractSyntaxTree.parse("-1")
=> (SCOPE@1:0-1:2 tbl: [] args: nil body: (INTEGER@1:1-1:2 -1))
irb(main):008>
``` -
-
- The expected value should be first, second the actual value.
-
05:39 AM Revision c8c0af3c (git): [ruby/rubygems] Fix grammar in gem installer cache comment
- Change "and lock the mutex" to "and to lock the mutex" for correct
parallel infinitive structure.
https://github.com/ruby/rubygems/commit/94f9267b72
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> -
- be checks object identity which will fail for newly-allocated strings.
eq checks value equality and is the correct matcher here.
https://github.com/ruby/rubygems/commit/828440937b
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@a... -
- The Rust function hello("world") returns "Hello world, from Rust!" but
the Ruby test templates expected "Hello earth, from Rust!", causing
generated tests to fail immediately after bundle gem --ext=rust.
https://github.com/ruby/rubygems... -
- https://github.com/ruby/rubygems/commit/72bdf54b6d
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> - 05:05 AM Revision b11091ca (git): [ruby/rubygems] Use distinct gemspec filenames in stub_with_version and stub_without_version helpers
- Agent-Logs-Url: https://github.com/ruby/rubygems/sessions/4028db0e-e050-48af-9704-4219653a4753
https://github.com/ruby/rubygems/commit/951ef62f76
Co-authored-by: hsbt <12301+hsbt@users.noreply.github.com> - 05:05 AM Revision fdc2a611 (git): [ruby/rubygems] Fix gemspec filename inconsistencies in stub_with_version and stub_without_version
- Agent-Logs-Url: https://github.com/ruby/rubygems/sessions/8de78edb-b0d3-4fd3-bdad-898184561abd
https://github.com/ruby/rubygems/commit/5a87b86515
Co-authored-by: hsbt <12301+hsbt@users.noreply.github.com> - 04:27 AM Revision 3f850711 (git): [ruby/rubygems] Fix typo in test description: 'to' -> 'do'
- Agent-Logs-Url: https://github.com/ruby/rubygems/sessions/9cf2fa2e-02a9-4dde-a833-8ad11974e1eb
https://github.com/ruby/rubygems/commit/5ac4c8400b
Co-authored-by: hsbt <12301+hsbt@users.noreply.github.com> - 04:24 AM Revision a0092d2a (git): [ruby/rubygems] Fix test_execute_allowed_push_host response message to match freebird gem
- Agent-Logs-Url: https://github.com/ruby/rubygems/sessions/61efd9ab-67d3-4ce2-b81d-4b6e8ef07f99
https://github.com/ruby/rubygems/commit/bf73b51866
Co-authored-by: hsbt <12301+hsbt@users.noreply.github.com> - 04:24 AM Revision 6d4e0064 (git): [ruby/rubygems] Fix lockfile DEPENDENCIES section to use 'rubocop' instead of 'parallel_tests'
- Agent-Logs-Url: https://github.com/ruby/rubygems/sessions/dd120552-e56f-4a0e-9143-ec483aa07bfc
https://github.com/ruby/rubygems/commit/56a98274c3
Co-authored-by: hsbt <12301+hsbt@users.noreply.github.com> - 03:52 AM Revision 76eb974b (git): [ruby/rubygems] Fix inconsistent indentation at line 481 in git_spec.rb
- Agent-Logs-Url: https://github.com/ruby/rubygems/sessions/a9efe3b4-99c9-4af2-9954-a65a2859edfc
https://github.com/ruby/rubygems/commit/3d4e90a355
Co-authored-by: hsbt <12301+hsbt@users.noreply.github.com> -
- 02:24 AM Revision 3e299ba8 (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 2.74.0 to 2.75.0
- [Release notes](https://github.com/taiki-e/i...
04/07/2026
-
10:11 PM Bug #21640: Core Pathname is missing 3 methods / is partially-defined
- This should be solved with https://github.com/ruby/ruby/pull/16358, great!
-
07:01 PM Revision f62bec1b (git): Fix sync for `syntax_suggest`
- It is a bit special in that is uses `spec` for tests. Since https://github.com/ruby/ruby/pull/15015 those did not
get synced anymore.
This fixes the mapping to what is was before and pulls in missing changes -
- I am curious about this gem and will give it a read over if you post the link here.
-
- I am hopeful that in the medium-long term, ZJIT will be able to "see through" and even elide general Ruby object allocations when possible. We have some more cleanup and feature work before that becomes possible (for example, a function ...
-
- I had a bit of trouble following what is going on.
Lexing produces tokens, this renames to reflect that.
https://github.com/ruby/syntax_suggest/commit/d3386ab6f4 - 04:28 PM Revision 3a5bb48c (git): Fix "attrest" typo (#16672)
-
- Also add tests for the other types which currently don't have one
https://github.com/ruby/syntax_suggest/commit/db2dca2def -
- https://github.com/ruby/syntax_suggest/commit/c6bd4b53e6
-
- Maybe ripper fixed it since then. But prism also doesn't have this problem.
https://github.com/ruby/syntax_suggest/commit/8bc383b2a4 -
- Allows to remove some compatibility code with ripper
https://github.com/ruby/syntax_suggest/commit/3f2e9a7657 -
- I want to start porting this to prism without the translation layer.
https://github.com/ruby/syntax_suggest/commit/540d4840ab -
10:24 AM Revision 2fda295c (git): [DOC] Doc for File::Stat
-
- Sharing middle substrings have been behind a compilation flags for
a very long time, and it's unclear if we'll ever make it the default.
However, we can still share middle substrings without breaking the
zero-terminated contract if the ... -
- Applied in changeset commit:git|13926009ba829f42037719b9a163a41a8d67abed.
----------
[Bug #21983] Make `&nil` distinguishable from nonexistent block arg -
-
- Any substring of a 7bit string will be 7bit.
-
- Use source_root.parent instead of hardcoded relative path to derive the
repository root, so the code remains correct if the support file is ever
moved. Also fix the example Gemfile source path in DEBUGGING.md to
reflect the new top-level... -
- The bundler tests were writing temporary files to bundler/tmp, but this
moves them to the repository's top-level tmp directory instead, which is
already covered by .gitignore. All references in CI workflows, SimpleCov
filters, rubocop ex... - 02:25 AM Revision 90924bc3 (git): Bump the github-actions group across 1 directory with 3 updates
- Bumps the github-actions group with 3 updates in the / directory: [ruby/setup-ruby](https://github.com/ruby/setup-ruby), [lewagon/wait-on-check-action](https://github.com/lewagon/wait-on-check-action) and [taiki-e/install-action](https:/...
-
12:17 AM Revision fa8c9194 (git): ZJIT: Profiling: Separate bits for T_CLASS and T_MODULE
- So I don't need to use is_subclass_of(). We already have two branches
for the two types during profiling anyways. -
- Previously, we had none-sensical lines such as:
```
v15:CPtr[CPtr(0x1008)] = Const CPtr(0x1010)
```
Which was a contradiction as the pointer in the type specialization
should match the pointer on the right hand side. -
- It's likely some sort of mask or flags. Hexadecimal is more compact.
-
- This is a8f3c34556 ("ZJIT: Add missing guard on ivar access on
T_{DATA,CLASS,MODULE}") but for the polymorphic implementation in HIR
build.
04/06/2026
-
10:36 PM Revision a2a69b42 (git): Add assertion to rb_class_alloc
- We should only ever call rb_class_alloc with a T_CLASS. Adding this
assert/assumption allows us to elide later checks for T_CLASS> -
- This should avoid us emitting the check for a special constant.
-
- 09:38 PM Revision 7687c4d6 (git): ZJIT: Trace compile events and invalidation events (#16666)
- Trace compiles with `--zjit-trace-compiles` and invalidation events with `--zjit-trace-invalidation`. See [sample trace](https://ui.perfetto.dev/#!/?url=https://bernsteinbear.com/tmp-perfetto/perfetto-24803.fxt) (in a moment, when the fi...
-
08:42 PM Revision 8dd73008 (git): ZJIT: Replace Option<i32> with Recompile enum for side exit recompilation (#16665)
-
-
-
05:53 PM Revision 16da5f23 (git): Revert "Double sweep byte budget to preserve heap 1 behavior"
- This reverts commit c617c5ec85ff69a5a8b13c56d51fcd234c00e1e2.
-
- Anchors the historical 2048/1024 slot counts on the 80-byte
heap instead of the 40-byte heap. This isolates whether the
major GC elimination seen in railsbench was caused by heap 1's
halved budget in the previous commit. -
- Larger slot pools are less heavily used, so a fixed slot count
over-services them relative to allocation pressure. Divide a
byte budget by heap->slot_size so the effective per-step slot
count tapers inversely with slot size. -
04:47 PM Bug #21983: RubyVM::AST no way to find &nil
- I am guessing that the `&nil` case should read `block: false` by comparing it to the `**nil` case:
```shell
❯ ruby -e 'pp RubyVM::AbstractSyntaxTree.parse("def foo(**nil); end")'
(SCOPE@1:0-1:19
tbl: []
args: nil
body:
(DEFN... -
- ```
$ ./ruby -e 'pp RubyVM::AbstractSyntaxTree.parse("def foo(&nil); end")'
(SCOPE@1:0-1:18
tbl: []
args: nil
body: (DEFN@1:0-1:18 mid: :foo body: (SCOPE@1:0-1:18 tbl: [] args: (ARGS@1:8-1:12 pre_num: 0 pre_init: nil opt: nil fir... -
- I don't have a particular opinion on this feature, but if the main concern is performance, I think `bigdecimal` could likely be optimized first. Of course having to allocate an object will always be slower than an immediate, but as menti...
-
05:59 AM Feature #21982: Add `Decimal` as a core numeric class
- What version of bigdecimal did you use for the benchmark?
We've released [bigdecimal v4.1.1](https://github.com/ruby/bigdecimal/releases/tag/v4.1.1) 2 days ago.
The calculation performance is improved in this version by supporting an e... -
- Mostly the same as for `on_kw`.
Stop comparing order for a handful of events. They are emitted by ripper in a order that is
not easy for prism to replicate. For example in`foo, bar, baz = 123`, `op` for `=` is emitted
before the last ma... -
- Extract the first element instead of `join.to_i` (`String#unpack1` is
available since ruby 2.4.0, but this gem still supports 2.3).
https://github.com/ruby/resolv/commit/f518da95ae -
- Limit the range from -100 km to the upper limit of the 32-bit
representation.
https://github.com/ruby/resolv/commit/e6dfbe7cb0 -
- `Size` smaller than 1m was represented as rational.
```ruby
Resolv::LOC::Size.create("0.01m").to_s #=> "1/100m"
```
Fix to represent in the proper form for `create`.
https://github.com/ruby/resolv/commit/2ff9e0585d -
- Limit the size to between 0cm and less than 100,000km.
https://github.com/ruby/resolv/commit/9d9e0c3044 -
- - Leading zeros in fractional part are significant, must not be
suppressed.
- Use `Integer#divmod` to calculate division and module at once.
- Use simple comparison for exact match than regexp.
https://github.com/ruby/resolv/commit/0e... -
- Limit the east-west direction to 180 degrees or less and the
north-south direction to 90 degrees or less, respectively.
https://github.com/ruby/resolv/commit/8075a1fe76 -
- Fix `Resolv::LOC::Coord.create` hemisphere handling for LOC
coordinates. `hemi` was always `1` because `m[4]` matched `([NESW])`.
https://github.com/ruby/resolv/commit/043b5bf624 -
- This method is available since ruby 1.9.0.
https://github.com/ruby/resolv/commit/1e05d06fbd -
- https://github.com/ruby/resolv/commit/451d3f367c
-
09:34 AM Revision 9dab3c51 (git): [DOC] Mention `Hash#default_proc=` affects Proc object from `Hash#to_proc`
- ```ruby
h = {foo: 0, bar: 1, baz: 2}
proc = h.to_proc
proc.call(:nosuch) # => nil
h.default_proc = proc {'wow!'} # This affects `proc` immediately
proc.call(:nosuch) # => 'wow!'
```
This behavior is a bit of surprise to me, and I could ... - 09:23 AM Revision e0a7dad9 (git): Update bundled gems list as of 2026-04-06
-
09:22 AM Revision ed6e7006 (git): Pin RBS to avoid bigdecimal validation failure (#16664)
-
- Using `next` in `-p` mode skips the implicit `print $_`, causing
pinned gem lines (e.g. rbs) to be silently dropped from the output.
Replace `next` with an `unless` guard so the line is always printed.
Co-Authored-By: Claude Opus 4.6 (1... - 02:30 AM Revision 3ef41a52 (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 2.71.2 to 2.73.0
- [Release notes](https://github.com/taiki-e/i... -
04/05/2026
-
07:39 PM Feature #21869: Add receive_all Method to Ractor API for Message Batching
- This is my first pr and issue. I’m unsure about the current status — is the request rejected, or are there any actions required from my side?
-
-
- The implications of that are, assuming this proposal is implemented based on start line/column + end line/column (or equivalently, start & end offsets):
* No need for `node_id` to implement this feature
* This feature works with both `... -
- I thought more about `node_id` and I found at least one case where it is problematic with different versions of Prism.
The problem is the `node_id` from the bytecode is computed by the builtin Prism parser used by `prism_compile.c`, whil... -
-
-
01:21 PM Feature #21982: Add `Decimal` as a core numeric class
- naruse (Yui NARUSE) wrote in #note-2:
> Combining two different precision decimals doesn't sound reasonable.
> ...
I think that's probably fair. I was thinking Fixnum/Flonum and their heap counterparts, but that's backing the same prec... -
- Eregon (Benoit Daloze) wrote in #note-1:
> For context, how much of this is AI-generated?
I implemented the i128 fixed-point arithmetic by porting Roc's Dec, then used AI to explore Ruby-specific optimizations (interpreter fast paths, Y... -
01:00 PM Feature #21982: Add `Decimal` as a core numeric class
- Combining two different precision decimals doesn't sound reasonable.
It should use a single precision to ease to understand how it works.
We long discussed about decimal including introducing based on C _Decimal128, but I don't under... -
- For context, how much of this is AI-generated?
Is it necessary to be faster than BigDecimal, for which realistic use cases?
Could BigDecimal be improved instead?
I think the bar for a new numeric core type is very high, and this w... -
- No need to list them all out anymore, it's just a single possibility now
https://github.com/ruby/prism/commit/893f2734cb -
- It can only ever be one of these two
https://github.com/ruby/prism/commit/f871ce0260 -
-
05:13 AM Feature #21704: Expose rb_process_status_new to C extensions
- @akr
> Why Process::Status object is created in such situations?
Because `Process::Status` represents either "the result of waitpid" or "errno of waitpid" and is used to transfer this information until it's finally consumed by the end ...
04/04/2026
-
- # Feature: Add `Decimal` as a core numeric class
## Abstract
Add `Decimal < Numeric` to Ruby core: exact base-10 arithmetic using a tagged immediate VALUE (like Fixnum) for small values, promoting to a 128-bit heap object for large... -
- Those FDs still live in the kernel space but have been closed in the
user space. -
- ruby_4_0 already contains commit:c353b625297162024b5a80480664e599dd49a294 and commit:f430fbbfacea5690d790dd9060ca4118431fc2fb.
-
- ruby_4_0 commit:7ba583d753411a39dec78ea6e7c8d2701b04c118 merged revision(s) commit:c27ae8d91aadca0660070ee1eeae9598b1fe47ee, commit:b6e6ccc6c22ebb464ce101e42a14e7daea2a80b6, commit:cbca59377ebfa1bc1183322fa9d9d0067d445c24.
-
- [ruby/psych] Remove excessive check of message
Make `Data#initialize` reject `Integer` keys
[Bug #21844] Fix error message -
- ruby_4_0 commit:2b22593ac1 already merged.
-
- ruby_4_0 commit:fda2bc4ee3045a17ca52ca6075cdd955a813fbc1 merged revision(s) commit:8ecf28f9384207590be73d5f2a95859cff9c526a.
-
- [ruby/rubygems] Fix NoMethodError in Gem.try_activate when activation conflicts occur
-
- ruby_4_0 commit:c38f8732c4ae6448cf05c795ddd5df4040ceeea8 merged revision(s) commit:c8155822c460a5734d700cd468d306ca03b44ce4.
-
- reinit rb_internal_thread_event_hooks_rw_lock at fork
-
- ruby_4_0 commit:4fa4d71d751a4efeb5616ca7227fa57df0e718b0 merged revision(s) commit:54c4694994cc3bcfea9058b22ba3e68af6aaf740.
-
- marshal.c: properly freeze linked strings
-
- ruby_4_0 commit:6b7f5aa73199616fd438847059187a95c0c6ef0f merged revision(s) commit:5de4cc56086493689701e86aa0ccf6a4a4a87d75.
-
- Fix regexp performance regression for patterns starting with s/k
04/03/2026
-
- Address review feedback: use a method instead of directly comparing
.status == IseqStatus::Invalidated. -
-
- When two GuardType instructions in the same basic block guard the same
value with the same (or narrower) type, the second guard is redundant.
For example, `(n - 1) + (n - 2)` generates two `GuardType n, Fixnum`
— one for each subtraction... -
-
-
-
- 02:23 AM Revision 480d0375 (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 2.71.1 to 2.71.2
- [Release notes](https://github.com/taiki-e/i... -
- When a class or module is cloned, Ruby currently rewrites the CREF chain of each copied method so that it points at the new class instead of the original. I don't think this should happen, and methods on a cloned class should retain thei...
-
-
-
- This reverts commit 33b9d2ab9f45a0c66665fb8dc40ddebd07c7b7b7.
-
- This reverts commit d368d42ec6b50cece947556a2185ce73fa493bfe.
-
- Despite adding rb_ec_stack_check() to Rust code entry points, we've seen
SystemStackError causing test failures due to debug builds consume too
much native stack space. Let's use opt-level=1 so rustc/LLVM are more
efficient with stack sp... -
- The transmute triggered a miscompilation on Rust 1.85.0, where the
callback was "optimized" to unconditional go to a `SIGTRAP`ping
instruction.
While later Rust versions don't have the same miscompilation, it seems
that this is also par...
04/02/2026
-
- Reduce the number of spawned processes from 50 to 20 and increase the
timeout from 30s to 60s. Process creation is expensive on Windows, and
the combination caused the test to exceed the 30s limit on Windows CI. -
- The default limit of 2.0 is too tight for RSS measurements on macOS,
causing random CI failures when the ratio lands just above 2.0 (e.g.
2.05). Increase to 2.2 to account for normal RSS measurement noise.
https://github.com/ruby/ruby/a... -
- warn_unused_block uses the caller's PC to build a dedup key for the
unused block warning table. When called from ZJIT-compiled code via
rb_vm_send, cfp->pc may be stale (poisoned or uninitialized) because
ZJIT stores the real PC in cfp->... -
-
- The server thread writes the HTTP response while the client has a
0.1s read_timeout. On slow CI (especially Windows), the client can
time out and close the socket before the server finishes writing
headers or the response body, causing E... -
- `yes-update-default-gemspecs` uses `$(XRUBY)` which execs the ruby
binary, but only depended on `main` (extensions and encodings).
When `make runirb` runs with parallel jobs, the linker may still be
writing the ruby binary while `update-... -
- The PR has been upstreamed.
-
-
- When there is a BOM on a file, the offsets are incorrect for nodes. See:
```
RubyVM::AbstractSyntaxTree.parse("\xEF\xBB\xBF1").source # => "\xEF"
```
-
- Sorry to be late: https://github.com/nobu/bashism
-
11:56 AM Bug #21933: Ruby::Box: named capture local variable can become nil after non-matching lines
- I opened a pull request with a fix and regression test for this issue:
https://github.com/ruby/ruby/pull/16644
-
-
- Applied in changeset commit:git|4009b713a64f469e593ffbc5bb4b37eff43f3be3.
----------
[Bug #21974] Fix RubyVM::AST inspection for ::Foo = 1 (#16642)
* Fix AST CDECL children for top-level constants
* Simplify cdecl AST regression test -
- I have created a fix for this https://github.com/ruby/ruby/pull/16642
-
- * Fix AST CDECL children for top-level constants
* Simplify cdecl AST regression test - 05:36 AM Revision 3319be01 (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 2.70.4 to 2.71.1
- [Release notes](https://github.com/taiki-e/i... -
-
- https://bugs.ruby-lang.org/issues/18254#note-14
I think `String#unpack`/`String#unpack1` should allow negative offset as well as the other methods, for example `String#slice` that is referred to in the description.
[ruby/ruby#16641]... -
- I wasn't aware that negative offset is not allowed.
Shouldn't `unpack`/`unpack1` also allow it as well as the other methods, `String#slice`, `String#[]` and so on? -
- Applied in changeset commit:git|91ae69860591f0dfc85571cfbd023dd199e17e73.
----------
Use compile-time flag to indicate dynamic CREFs
The inline constant cache previously used `RCLASS_SINGLETON_P` to detect
"unstable" CREFs that need ic... -
- The inline constant cache previously used `RCLASS_SINGLETON_P` to detect
"unstable" CREFs that need ic_cref stored and checked on every IC hit.
This caused the `class << self` pattern to create inline caches which
requires extra checks a... -
- `String#unpack` does not allow negative offset.
- 03:17 AM Revision 81df2bda (git): [ruby/rubygems] Bump the rb-sys group across 2 directories with 1 update
- Bumps the rb-sys group with 1 update in the /test/rubygems/test_gem_ext_cargo_builder/custom_name/ext/custom_name_lib directory: [rb-sys](https://github.com/oxidize-rb/rb-sys).
Bumps the rb-sys group with 1 update in the /test/rubygems/t... -
- When no gems have missing extensions (e.g. on JRuby where
missing_extensions? always returns false), display an informational
message instead of raising Gem::Exception.
https://github.com/ruby/rubygems/commit/738d2dbdf6
Co-Authored-By:... -
- JRuby does not require extension rebuilds when switching JRuby or Java
versions because Java extensions are shipped pre-compiled and JRuby does
not expose a version-specific C API. This means missing_extensions?
returning true on JRuby i... -
- Replace gzio.read with IO.copy_stream(gzio, IO::NULL) to avoid
allocating a potentially large string when discarding unconsumed
compressed data.
https://github.com/ruby/rubygems/commit/e4d0196887
Co-Authored-By: Claude Opus 4.6 (1M con... -
- When a GzipReader is closed before all compressed data is consumed,
zlib emits "attempt to close unfinished zstream; reset forced" warning.
This happens in open_tar_gz because the TarReader stops reading at the
end-of-archive marker, lea... -
- Xcode 26.5 beta ships a clang that promotes -Wincompatible-function-pointer-types to an error by default. The test's inline C extension defined `VALUE foo()` (taking no arguments) but passed it to `rb_define_global_function` which expect...
-
- Applied in changeset commit:git|dd3542a0cd00b40d406763e85ac12a1086bab682.
----------
Fix class variable cache not invalidated by class_variable_set
The previous invalidation walked subclasses but missed cvars from
included modules, and... -
- The previous invalidation walked subclasses but missed cvars from
included modules, and skipped invalidation when creating a new cvar on
modules entirely. Always invalidate when a new class variable is
created, since this should be relat... -
- This reverts commit 77b803cfc833233a49b497a9342f16c9237833e6.
There is a new NULL dereference in tool/test/testunit/test_sorting.rb
that seem to have started with this commit.