Repository navigation
ci: ci-ok runs unless the run was cancelled (port #111 to main) - #114
Merged
Merged
Conversation
#111) always() also ran it on a run that concurrency cancel-in-progress had cancelled, and it failed on the cancelled jobs: a red check for a commit nobody was judging any more (taylab-launch-kit v1.40.4). It still runs when a job fails or times out. (cherry picked from commit 56cb295) Ported to main from develop: main's ci-ok gate still used always(), and 18 of the 25 ci-ok failures 10-11..10-21 were runs that cancel-in-progress had cancelled. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019jHjgRysB6GajV3oDppruf
3 of 4 tasks
addisdev
pushed a commit
that referenced
this pull request
Oct 26, 2026
Brings back what landed on main without going through develop: #112 (Node.js 24 action majors), #114 (already here as #111; identical), #110 vite 7.3.7, #109 fastify 5.12.6, #113 fast-uri, and #116 (gradlew.bat and OFL-IBMPlex.txt stored with LF). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019jHjgRysB6GajV3oDppruf
5 of 6 tasks
addisdev
pushed a commit
that referenced
this pull request
Oct 26, 2026
ubuntu-latest has started handing out ubuntu-26.04 images, where
webkit2gtk-driver is gone ("no installation candidate"; replaced by
webkitgtk-webdriver, which still ships /usr/bin/WebKitWebDriver). The
desktop launch smoke failed at apt-get on PR #117's first run; #114's
green run the day before was on ubuntu-24.04.
Pick the package by whether apt has a candidate for the new name:
`apt-cache show` succeeds on 24.04 too, which knows the name and has
nothing to install under it. Checked against both releases' indexes:
24.04 picks webkit2gtk-driver 2.52.6, 26.04 picks webkitgtk-webdriver
2.52.0, and every other package in the step has a 26.04 candidate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019jHjgRysB6GajV3oDppruf
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 join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
What and why
This is a verbatim cherry-pick of #111 (
56cb295, merged intodevelopon 10-24) ontomain. It changes one line in.github/workflows/ci.yml:ci-ok'sif: always()becomesif: ${{ !cancelled() }}. The patch is identical todevelop's, so the nextdevelop→mainrelease merges cleanly.Why main needs it now.
ci-okfailed 25 times between 10-11 and 10-21 (every attempt of everycirun, read from the Actions API). Those 25 failures are spread over 13 distinct branches. In 18 of the 25, no job failed. In each of those,concurrency: cancel-in-progresscancelled the run because a newer push had arrived.always()still ranci-ok, which sawcancelledand went red on a commit nobody was judging any more. This is the only cause that recurs; every other cause appears once.cancelledalways()binaries/fleet-…not a directory;EACCES mkdir /home/fleet/.fleet/datafleet/src/service.tschange/bin/bash(ENOENTon Windows)EBUSYunlinkingfleet.dbin test cleanupfleet/test/helpers.tsretries the delete)SocketExceptionfetching android-all from Maven; attempt 2 passedisReleasePrinscripts/changelog-coverage.mjs)Verified
ci.ymlparses; the diff againstdevelop's copy of the job is empty.what changedhad finished.ci-okconcludedcancelled, notfailureand notskipped; theci-okcheck run on7628d16readcancelledtoo. A cancelled required check does not count as passing, so a run cancelled by hand still blocks the merge, asalways()did.ci.ymlchange runs every suite), from a rerun of that run.CHANGELOG.md: not user-visible (CI only), same as ci: ci-ok runs unless the run was cancelled #111.What is still not true
cancelledci-okis still a non-green mark on the superseded commit. That's intended (it is notfailure), but any tool that counts every non-success as red will still count it.ci-ok.🤖 Generated with Claude Code
https://claude.ai/code/session_019jHjgRysB6GajV3oDppruf