ci: Lint follow-ups

maflcko commented at 1:30 pm on November 4, 2025: member

This contains a few follow-ups to #33744:

Rewrite the actions Bash snippet to Python. I’ve confirmed it still works in https://github.com/maflcko/bitcoin-core-with-ci/actions/runs/19067932430 (scroll down).
Add a lint-build retry to avoid issues such as #33640 for the lint task as well.
Finally, run the debug_pull_request_number_str annotation on all checks, to ensure they are present even when GitHub deletes annotations on a re-run. For example, the initial attempt https://github.com/bitcoin/bitcoin/actions/runs/19041534107/attempts/1?pr=33772 has the annotations, and the lint re-run has them removed: https://github.com/bitcoin/bitcoin/actions/runs/19041534107?pr=33772

DrahtBot added the label Tests on Nov 4, 2025

DrahtBot commented at 1:30 pm on November 4, 2025: contributor

The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

Code Coverage & Benchmarks

For details see: https://corecheck.dev/bitcoin/bitcoin/pulls/33776.

Reviews

See the guideline for information on the review process.

Type	Reviewers
ACK	m3dwards, willcl-ark

If your review is incorrectly listed, please react with 👎 to this comment and the bot will ignore it on the next update.

maflcko force-pushed on Nov 4, 2025

DrahtBot added the label CI failed on Nov 4, 2025

fanquake commented at 1:57 pm on November 4, 2025: member

cc @willcl-ark @m3dwards

DrahtBot removed the label CI failed on Nov 4, 2025

maflcko force-pushed on Nov 7, 2025

ci: Rewrite Bash to check inputs to Python

This is shorter and easier to read. Also, according to the dev notes,
Bash should not be used.

fa0d37a579

ci: Rewrite lint task Bash snippet to Python

The Bash snippet was shorter, but relying on implicit word splitting
(see the shellcheck SC2086 warning).

For example, the DOCKER_BUILD_CACHE_ARG shlex.split is now done
identical to how ci/test/02_run_container.py does it.

Moreover, the Python will hopefully be easier to modify in the future,
as the dev notes recommend Python over Bash.

fac4f6de28

maflcko force-pushed on Nov 12, 2025

in .github/workflows/ci.yml:641 in fa213dc40b outdated

643+              "."
644+          ]
645+
646+          if run(build_cmd, check=False).returncode != 0:
647+              print(f"Retry building image tag after failure")
648+              time.sleep(3)

m3dwards commented at 9:40 am on November 13, 2025:

Missing import time?

in .github/workflows/ci.yml:642 in fa213dc40b

644+          ]
645+
646+          if run(build_cmd, check=False).returncode != 0:
647+              print(f"Retry building image tag after failure")
648+              time.sleep(3)
649+              run(cmd_build)

m3dwards commented at 9:42 am on November 13, 2025:

Should be build_cmd

m3dwards commented at 9:54 am on November 13, 2025: contributor

The GHA behaviour of removing annotations from previous runs is very annoying! I experimented with having an upstream job that printed the annotation and having all the other jobs depend on it but again when you rerun a failed downstream job you lose the annotation from the run upstream job.

It’s a shame that in this PR we have to have the annotation printed many times, one per job, as I think it might drown out other notices. Without knowing exactly how this machine readable annotation is being consumed or experimenting more, I’m not aware of an alternative? I considered storing it as an artefact and having each job check the artefact but I don’t think that works and is also ugly. Third thing I briefly considered is if the PR number could be output as it’s own check run somehow? Probably best to just leave it as it is here!

ci: Retry lint image building once after failure

The same was done for the other CI tasks in commit fa6aa9f42fa. This may
guard against intermittent network issues to download the base image or
packages ...

faf05d637d

maflcko force-pushed on Nov 13, 2025

maflcko commented at 10:52 am on November 13, 2025: member

It’s a shame that in this PR we have to have the annotation printed many times, one per job, as I think it might drown out other notices.

Agree, but I think we can use warning notices to bubble up more important ones.

Without knowing exactly how this machine readable annotation is being consumed or experimenting more, I’m not aware of an alternative?

It can be used via the API, e.g. https://github.com/maflcko/DrahtBot/blob/6c91378116516ca5974d78a4e0c5c495b38acf95/webhook_features/src/features/ci_status.rs#L84-L106

I’ve tried other things, such as task summaries (https://github.com/maflcko/DrahtBot/actions/runs/19115117072?pr=61#summary-54622116294) However, they are not really documented and can’t be retrieved via the API, IIRC.

Third thing I briefly considered is if the PR number could be output as it’s own check run somehow? Probably best to just leave it as it is here!

In theory we could hook up Cirrus again (or our own CI “dummy” provider), just to have a single task run via it, to annotate it with the pull number, but this seems more fragile.

Missing import time? …

Thx, fixed both. I guess this is the common reminder that a yaml string is the wrong place to put code. All ci code, even snippets, should be in a native file.

ci: Annotate all check runs with the pull request number

On check re-runs the annotations are discarded, so all check runs
require the number to be set.

fae3618fd6

maflcko force-pushed on Nov 13, 2025

DrahtBot added the label CI failed on Nov 13, 2025

DrahtBot removed the label CI failed on Nov 13, 2025

m3dwards commented at 3:52 pm on November 13, 2025: contributor

Looks good!

ACK fae3618fd6c82dfcea2f296caa16a79182b32059

willcl-ark approved

willcl-ark commented at 12:04 pm on November 17, 2025: member

ACK fae3618fd6c82dfcea2f296caa16a79182b32059

This all seems fine, and I agree python is superior to bash where possible. It is indeed annoying that GH doesn’t persist the annotations, but after also checking myself I can’t think of any better way to do it either.

fanquake merged this on Nov 17, 2025

fanquake closed this on Nov 17, 2025

maflcko deleted the branch on Nov 17, 2025

ci: Lint follow-ups #33776

Code Coverage & Benchmarks

Reviews