Skip to content

[pull] master from ruby:master#1055

Merged
pull[bot] merged 14 commits into
turkdevops:masterfrom
ruby:master
May 28, 2026
Merged

[pull] master from ruby:master#1055
pull[bot] merged 14 commits into
turkdevops:masterfrom
ruby:master

Conversation

@pull
Copy link
Copy Markdown

@pull pull Bot commented May 28, 2026

See Commits and Changes for more details.


Created by pull[bot] (v2.0.0-alpha.4)

Can you help keep this open source service alive? 💖 Please sponsor : )

tekknolagi and others added 14 commits May 27, 2026 15:07
We don't need to first turn a CBool into a Ruby bool then check that bit
pattern; we can go straight to working with the CBool.
JRuby's Marshal.dump emits an extra ivar for a compare_by_identity
Hash, which trips SafeMarshal's UnpermittedIvarError. Mirror the
existing pend used for test_time_with_ivar so the JRuby job stays
green.

ruby/rubygems@949cf8c4dc
warbler 2.0.5 (from the github pull/557 pin) does not work on JRuby
10.1: it cannot require the now-bundled ostruct, and even after that
hurdle Monitor#synchronize raises ArgumentError under the warble jar
build. warbler 2.1.0 explicitly supports JRuby 10.1 and pulls ostruct
in itself, so switch the fixture over to the gem release.

ruby/rubygems@c52563d372
Extend symlink_supported? to probe readlink so the symlink-based
tests omit cleanly when JRuby on Windows lacks readlink, switch
test_gem_paths_do_not_contain_symlinks to use the helper, and add
per-test omits for the remaining JRuby/Windows incompatibilities
(install permissions, project sanity, require/realworld, open
command, git submodule checkout, operating-system defaults).
test_self_load_escape_* now also rescues Errno::EACCES which JRuby
on Windows raises instead of Errno::EINVAL.

ruby/rubygems@321782d903
assert_self_install_permissions' ensure block calls File.chmod with
a splatted Dir.glob result; when the omit short-circuits the body,
the glob is empty and JRuby (unlike CRuby) rejects a chmod with no
paths. Guard the ensure to skip when no files were created.

test_generate_bin_symlink_win32 and test_generate_bin_with_dangling_symlink
fall into the "Unable to use symlinks" wrapper branch only when
symlink_supported? is false, but the installer on JRuby/Windows still
manages to create the symlink, so the expected warning never appears.
Omit them on the combo.

ruby/rubygems@fd318f75a6
…dows in tests

JRuby 10.1 on Windows raises TypeError inside File.symlink (wincode
trips on a nil path), so any test that exercises Gem::Installer's
symlink branch fails to even install the gem. Real users avoid the
branch entirely because `gem install` goes through DependencyInstaller
which passes wrappers: true. Mirror that production default for the
test suite by monkey-patching Gem::Installer.at and .for_spec in
test helper.rb so every direct caller (including the many tests that
bypass the installer_test_case helpers) gets wrappers: true on the
JRuby/Windows combo. The few tests that explicitly want the symlink
branch already set `installer.wrappers = false`, which still
overrides.

ruby/rubygems@88f5e29e62
…on Windows

Gem::Package#create_symlink already falls back to FileUtils.cp_r on
Windows when symlink creation hits Errno::EACCES (Developer Mode off),
but JRuby 10.1 surfaces the same failure as TypeError from its
internal wincode path helper. Treat TypeError as the same recoverable
condition so JRuby/Windows users extracting a tarball that ships
symlinks get the copy fallback rather than a crash.

ruby/rubygems@2d76d7bb73
This allows us to treat the two types of T_DATA the same way when
reading instance variables.

Non-TypedData T_DATA are deprecated, so making them faster isn't
particularly important, but it is helpful to avoid needing to test for
TypedData and adding another special case.
…ne (#17128)

In optimize_invokesuper's variadic (-1) arm, emit_super_call_guards was
called once before the inline attempt and again inside the successful
inline branch. The second call emitted a redundant GetEP/LoadField/Guard
chain (same level, same fields, same expected values) that the first
call already produced. The fixed-argc arm only called it once.

Drop the duplicate.
@pull pull Bot locked and limited conversation to collaborators May 28, 2026
@pull pull Bot added the ⤵️ pull label May 28, 2026
@pull pull Bot merged commit de6597d into turkdevops:master May 28, 2026
1 of 3 checks passed
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants