Hacktoberfest 2026:メンテナが10月に向けて印を付けた、オープンで初心者向けの issue。 Hacktoberfest の issue を見る

CI: a PR influx is serialised by 'require up to date' plus a 12 minute CI; consider a merge queue

オープン
#758 コメント 0 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

まだ誰も着手していません。

評価

難易度
5/5
見積もり時間
1週間以上
初心者へのやさしさ
15/100
issue の種類
機能追加
明瞭さ
おおむね明確
活発さ
活発
技術スタック
github, github-actions
領域
ci-cd

調査の方向性

Start with the on: trigger block in .github/workflows/ci.yml, which needs a merge_group entry, and with the branch protection settings for main, where strict and the merge queue are configured. Maintainers must make the settings change by hand, so a newcomer's part is mainly the workflow edit and the AGENTS.md note. Done means the chosen option is applied and a test merge goes through the queue, or the maintainers record a decision in the thread.

索引モデルが issue の本文から書いたものです。

説明

tech-debt

Merging 22 PRs on 2026-10-13/14 took about 15 hours of wall time, almost all of it waiting.

main has legacy branch protection with the required check All tests passed and strict: true ("require branches to be up to date before merging"). Every merge makes every other open PR stale, so each one has to be updated and re-tested (CI is 8 to 17 minutes, 12 on average, across a 15 job matrix) before it can merge, and only one PR can be in that state at a time. Roughly 12 minutes of CI times N pending PRs, serial.

Related costs seen the same day:

  • an update to a contributor's branch re-triggers GitHub's first-time-contributor approval gate for CI, so each cycle also needs a manual approval
  • a PR from a fork without "Allow edits from maintainers" cannot be updated by a maintainer at all (#719), so it waits for the author
  • retargeting a stacked PR after its base is squash-merged conflicts (add/add on the squashed file) and needs a git rebase --onto, then a close/reopen to start CI, because the workflow does not run on a base change

Options

  • GitHub merge queue: tests the PR on top of the queue head, so PRs need not be updated by hand and the queue runs them in order. Needs a merge_group trigger in .github/workflows/ci.yml and the queue enabled on main. The right fix
  • drop strict but keep the required check, accepting a possible untested combination; mitigated by CI running on main after each merge (it already does, on push)
  • if either is done, say so in AGENTS.md so tools stop assuming serial merges

A one-off relaxation of strict for a batch whose combination has been simulated is also an option, but was declined by the permission classifier as a CI bypass this time, so it has to be done by hand in the repo settings.

🤖 Generated with Claude Code

主要言語
C++
スター
3.5k
フォーク
630
平均マージ
2日 14時間
マージ済み PR(30日)
65

環境構築

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

petercorke/robotics-toolbox-python のほかの issue

petercorke/robotics-toolbox-python の issue をすべて見る

似ている issue

C++ の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。