[pull] master from ruby:master#1119
Merged
Merged
Conversation
When `limit` is negative, the trailing empty strings are included.
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 2.81.10 to 2.81.11 - [Release notes](https://github.com/taiki-e/install-action/releases) - [Changelog](https://github.com/taiki-e/install-action/blob/main/CHANGELOG.md) - [Commits](taiki-e/install-action@7a79fe8...15449e3) --- updated-dependencies: - dependency-name: taiki-e/install-action dependency-version: 2.81.11 dependency-type: direct:production update-type: version-update:semver-patch dependency-group: github-actions ... Signed-off-by: dependabot[bot] <support@github.com>
The hot thread cannot steal back control when sched->running is an MN
thread, and goes to sleep with sched->runnable_hot_th still pointing to
itself, firing VM_ASSERT(sched->runnable_hot_th != th) at the next
wakeup. Drop the hint before sleeping since the thread is no longer
spinning. On a VM_CHECK_MODE build with RUBY_MN_THREADS=1:
Thread.new { loop {} }
100.times { File.read(IO::NULL) }
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Fix ruby/rubygems#9170 - In #9131, we added running `make` in parallel with the `-j` flag. It's problematic because since Bundler installs gems in parallel, we could end up installing `N gem x M processors` which would exhaust the machine resources and causes issues like #9170. In this patch, I have implemented a make [jobserver](https://www.gnu.org/software/make/manual/html_node/POSIX-Jobserver.html), so that each thread installing a gem can take available job slots from the pool. I also want to highlight that with this patch, we can still end up running 2 more jobs than what's available. The reason being that if a gem is waiting for slots to be available, then the whole `bundle install` takes much longer which defeats the whole purpose of the `make -j` optimization. So if there is no more slots available, we compile the extension without parallelization (we still use one cpu though). Example ---------- If `bundle install -j 6` is running, then we have 6 slots available. Each gem with native extension that gets installed may take up to *3* slots (see my reasoning on this number below). In the event where 3 gems with native extensions get installed, 2 gems will consume all 6 slots, leaving the third gem without any. Ultimately, this means that we could use *at maximum 2 more slots than what's available* (e.g. `bundle install -j 3` for 3 gems will use 5 slots). I chose `3` slots per `make` process because after installing multiple gems, I didn't see any benefit to adding more jobs. It's possible that some gems have many `make` recipes that could run more than 3 in parallel, but I don't think this is very frequent. ruby/rubygems@664a6ff9c2
IO.new defaults to read mode when no mode is given. On POSIX, Ruby inspects the descriptor's access mode and opens the write end of the jobserver pipe for writing anyway, but on Windows that inspection is not available, so releasing slots raised "IOError: not opened for writing". Pass explicit modes so the write end is usable on every platform. ruby/rubygems@9d17c614dd Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The guard that avoids appending a duplicate -j used /-j\d?(\s|\Z)/, but \d? matches at most one digit, so an existing -j10 (or any two-digit job count) failed to match and a second -j was appended anyway. Use \d* so a job count of any width is recognized. ruby/rubygems@2bb45e9cd4 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
connect_to_jobserver matched the first --jobserver-auth in MAKEFLAGS, but the value is appended after any pre-existing flags. When bundle install runs under a parent make jobserver, MAKEFLAGS already carries the parent's --jobserver-auth, so the leftmost match returned the parent's descriptors instead of the pool ParallelInstaller just created. Scan and take the last match so we always connect to our own pipe. ruby/rubygems@e42c907190 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
If the read end raised while closing, the original ensure skipped both the write-end close and the MAKEFLAGS restore, leaving --jobserver-auth with closed descriptors in the process environment. Restore MAKEFLAGS first and guard the closes so the environment is always cleaned up. Building the value with compact.join also drops the stray leading space when MAKEFLAGS was previously unset. ruby/rubygems@ae2530f941 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
redefine_build_jobs coordinated the two worker threads with `sleep(0.1) while flag` loops. If a thread died before clearing its flag the other spun forever, and the polling added latency to every run. Use blocking Thread::Queue handoffs so the ordering is deterministic and a failure surfaces instead of hanging on a flag that never flips. ruby/rubygems@3f6745222d Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Name the per-gem slot cap MAX_JOBS_PER_GEM with a comment explaining the limit, rename the local to acquired_jobs since it counts grabbed tokens rather than free ones, and rescue EOFError alongside IO::WaitReadable so a closed jobserver pipe falls back to a single job instead of raising. ruby/rubygems@746ffc2381 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Bundler's quality spec rejects weak modifiers in lib source, and the connect_to_jobserver comment used "the pool we just created". Name ParallelInstaller instead so the meaning stays clear without "just". ruby/rubygems@a9818a7119 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Fix: ruby/json#1008 I'm unfortunately unable to turn the reproduction script into a test case there, I don't understand what is causing `err_info` to be re-raised in the repro that doesn't work in test-unit. ```ruby require 'json' parser = JSON::ResumableParser.new({}) ['{"message": "hello ', 'world"}', '[1,2]'].each do |chunk| parser << chunk begin while parser.parse p parser.value end rescue JSON::ParserError => e p e parser.clear end end ``` ruby/json@24a2b083d7
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
5 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
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 : )