Stop using auto

[jwodder]

Closes #7016. See https://github.com/jwodder/pr-merge-release-test for a test of the workflow in action.

Closes #6856 , Closes #6918

TODOs:

• comments on released PRs & issues
• does not overwrite old release notes; see Running github-release only for the most recent release nedbat/scriv#57 -- was tested to work - we just need scriv release

[codecov]

Codecov Report

Base: 75.41% // Head: 76.21% // Increases project coverage by +0.79% 🎉

Coverage data is based on head (8b8446e) compared to base (9567368).
Patch has no changes to coverable lines.

❗ Current head 8b8446e differs from pull request most recent head 8d99113. Consider uploading reports for the commit 8d99113 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##            maint    #7024      +/-   ##
==========================================
+ Coverage   75.41%   76.21%   +0.79%     
==========================================
  Files         354      354              
  Lines       58751    58750       -1     
  Branches     6613     6613              
==========================================
+ Hits        44310    44779     +469     
+ Misses      14427    13957     -470     
  Partials       14       14              

Impacted Files	Coverage Δ
datalad/_version.py	45.68% <0.00%> (-0.28%)	⬇️
datalad/tests/test_utils.py	70.28% <0.00%> (+0.09%)	⬆️
datalad/tests/utils_pytest.py	90.12% <0.00%> (+0.09%)	⬆️
datalad/distributed/ora_remote.py	80.46% <0.00%> (+0.15%)	⬆️
datalad/local/tests/test_add_archive_content.py	74.16% <0.00%> (+0.27%)	⬆️
datalad/support/tests/test_sshrun.py	70.73% <0.00%> (+1.21%)	⬆️
datalad/cli/tests/test_main.py	63.49% <0.00%> (+1.37%)	⬆️
datalad/tests/utils.py	56.58% <0.00%> (+7.92%)	⬆️
datalad/tests/test_tests_utils.py	68.55% <0.00%> (+68.55%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[yarikoptic]

since we need now to wait for scriv to allow to work only on the target latest release for github (judging from author speed of reply it would be done in some foreseable future so we don't need to implement it ourselves), would you be able to add a helper for released auto plugin functionality, i.e. to annotate issues and PRs with the version it was fixed in?

[jwodder]

@yarikoptic Commenting-script added; you can see it in action in the latest PRs in the test repo linked above.

[yarikoptic]

ok, now we just need to wait for @nedbat to improve scriv (nedbat/scriv#57) and then we would merge/test-drive in datalad. Then IMHO we should centralize all this goodness into a GitHub action so it could be reused by datalad and its extensions at minimum. I have initiated https://github.com/datalad/release-action repository for that and made you @jwodder an admin for it happen you want to move this goodness there right now. Note that we have https://github.com/datalad/datalad-action already which could provide some hints on how github action could be developed.

[yarikoptic]

note: @jwodder, I have updated description to close more issue and become TODOs instead of shortcomings (already addressed)

[yarikoptic]

4 @jwodder and FTR, to mitigate the issue of changelog commit pushed from github actions I have added yarikoptic-gitmate user with Write permissions to this repo and added GITMATE_GITHUB_TOKEN secret so workflow could use that token instead of default one to push.

[jwodder]

@yarikoptic Do you want workflows to run on the changelog-creation commit? I thought you were concerned about running them on snippet-creation commits for the add-changelog-snippet.yml workflow.

[yarikoptic]

Would it not be safe to rerun add-changelog-snippet.yml workflow upon such a push because label would no longer be assigned and thus that workflow would just do nothing (no infinite recursion of commits to happen)?

[jwodder]

@yarikoptic The workflow won't run again once the "CHANGELOG-missing" label is removed, but there's a slight chance that, after the snippet is pushed, a new workflow run might start before the first workflow removes the label.

[yarikoptic]

good observation! My thinking is: we can then removed label before pushing the snippet. But then if snippet push fails - we would end up without the snippet and label, which would be odd... but probably it would be also as odd if now push of snippet fails and we do not have label removed. So IMHO it would be ok to just swap to steps in the order so we would completely avoid such a race condition. Am I missing any potential drawback/side-effect @jwodder?

[jwodder]

@yarikoptic I can't think of any drawbacks. Should I update add-changelog-snippet.yml in this PR or a separate one?

[yarikoptic]

@yarikoptic I can't think of any drawbacks. Should I update add-changelog-snippet.yml in this PR or a separate one?

I think you can do right in this one and we will merge it altogether to give it a shot.

[jwodder]

@yarikoptic Steps swapped. Do you also want GITMATE_GITHUB_TOKEN to be used in either or both workflows in this PR?

[yarikoptic]

@yarikoptic Steps swapped.

Thank you!

Do you also want GITMATE_GITHUB_TOKEN to be used in either or both workflows in this PR?

I don't care much but may be let's do for both for consistency?

[jwodder]

@yarikoptic Done: a73411a.

[yarikoptic]

Thank you @jwodder ! So I will proceed with merge and we will try it out

[yarikoptic]

FWIW new scriv https://pypi.org/project/scriv/0.17.0/ was released hours back and the change is

The github_release command now only considers the top-most entry in the changelog. You can use the --all option to continue the old behavior of making or updating GitHub releases for all of the entries.

so I do not think we need any tune up to this PR.

[yarikoptic-gitmate]

PR released in 0.17.6