git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH v5 0/2] ci: link failure and leak annotations to the test script

From
Harald Nordgren <haraldnordgren@gmail.com>
Date
Oct 4, 2026, 12:03 UTC
Message-ID
<CAHwyqnWU0z7wHxDnxdoviOBenYJogB7k4GLo2jdO7_M_hSMZWQ@mail.gmail.com>
In-Reply-To
<3587bf44-1b3b-422f-a926-f8481104dfd8@gmail.com>
On Sat, Oct 3, 2026 at 3:03 PM Phillip Wood <phillip.wood123@gmail.com> wrote:
Show 55 quoted lines
>
> Hi Harald
>
> On 03/10/2026 09:11, Harald Nordgren via GitGitGadget wrote:
> > Link failure and leak annotations in CI to the test script, so both can be
> > found from the job summary.
> >
> > V5 CI Job where failures and leaks are reported:
> > https://github.com/git/git/actions/runs/36979818141/job/110752027863?pr=2426
>
> There does not appear to be any output relating to leaks in that job.
> The first patch hasn't changed so I'm not sure why that is.
>
> > Changes in v5:
> >
> >   * Removed % escaping entirely, verified on CI that it isn't needed. Every
> >     existing test description that uses % renders correctly unescaped.
> >   * Rewrote the file/line commit message with a concrete example (failed:
> >     t1060.17 partial clone of corrupted repository).
>
> You have added
>
>
>      When a test fails, GitHub shows an annotation naming it, for
>      example:
>
>          failed: t1060.17 partial clone of corrupted repository
>
> which shows an example of the current output without the filename or
> line annotations. There is no example of what that output changes to, so
> there is no way for someone reading that message to see what has
> actually changed. After spending some time clicking around in Github I
> think what that patch changes is not the test output of individual jobs
> which you linked to above, but what is displayed on the summary page at
>
> https://github.com/git/git/actions/runs/36979818141?pr=2426
>
> That page shows a list of annotations with links to the changes in the
> failed test file. That is a useful improvement but how you expected
> someone reading the commit message to understand what had changed when
> you did not give an example of the new output, and the changes are on a
> different page to the one you linked to in the cover letter is beyond
> me. I'm pretty exasperated that I've had to spend time messing about on
> Github trying to see what has changed because you could not provide a
> link and write a couple of sentences explaining it. After asking what
> this change did in v3 you replied that the commit message wasn't clear
> without explaining what the change actually did. When I asked what the
> change did in practical terms in response to v4 I got no reply. As you
> already know reviewer time is short on this list, so please, when
> someone asks a question answer it rather than replying with an obtuse
> comment or simply ignoring it and sending another patch.
>
> Both these patches are useful improvements, but trying to get an
> explanation of what they did has been like trying getting blood out of a
> stone.
I don't understand why this tone is necessary at all.
Yes, I should clarify that it affects the summary.

But I posted a comment regarding the regression you brought up in your following message, so we should decide what we want before progressing. The point of this whole topic was to make the leak reporting less bad, everything else was a bonus.

Harald
Previous: Phillip WoodNext: Harald Nordgren via GitGitGadget
Message 36 of 40 in “ci: point leak-sanitizer failures at the actual test and error”
  1. ci: point leak-sanitizer failures at the actual test and errorHarald Nordgren via GitGitGadget, Sep 25, 2026
  2. Ben KnobleSep 25, 2026
  3. Harald NordgrenSep 25, 2026
  4. Phillip WoodSep 27, 2026
  5. Harald NordgrenSep 27, 2026
  6. 0/2 ci: link failure and leak annotations to the test scriptHarald Nordgren via GitGitGadget, Sep 28, 2026
  7. 2/2 ci: point test failures and fixed known breakages at their file and lineHarald Nordgren via GitGitGadget, Sep 28, 2026
  8. Junio C HamanoSep 28, 2026
  9. 1/2 ci: annotate leaks and stop a leak-sanitizer script at its first failureHarald Nordgren via GitGitGadget, Sep 28, 2026
  10. Junio C HamanoSep 28, 2026
  11. Harald NordgrenSep 29, 2026
  12. 0/2 ci: link failure and leak annotations to the test scriptHarald Nordgren via GitGitGadget, Sep 30, 2026
  13. 1/2 ci: annotate leaks and stop a leak-sanitizer script at its first failureHarald Nordgren via GitGitGadget, Sep 30, 2026
  14. Phillip WoodSep 30, 2026
  15. 2/2 ci: point test failures and fixed known breakages at their file and lineHarald Nordgren via GitGitGadget, Sep 30, 2026
  16. Phillip WoodSep 30, 2026
  17. Harald NordgrenSep 30, 2026
  18. Junio C HamanoSep 30, 2026
  19. Phillip WoodSep 30, 2026
  20. Junio C HamanoSep 30, 2026
  21. 0/2 ci: link failure and leak annotations to the test scriptHarald Nordgren via GitGitGadget, Oct 1, 2026
  22. 1/2 ci: annotate leaks and stop a leak-sanitizer script at its first failureHarald Nordgren via GitGitGadget, Oct 1, 2026
  23. 2/2 ci: point test failures and fixed known breakages at their file and lineHarald Nordgren via GitGitGadget, Oct 1, 2026
  24. Phillip WoodOct 1, 2026
  25. Junio C HamanoOct 1, 2026
  26. Harald NordgrenOct 2, 2026
  27. 0/2 ci: link failure and leak annotations to the test scriptHarald Nordgren via GitGitGadget, Oct 3, 2026
  28. 1/2 ci: annotate leaks and stop a leak-sanitizer script at its first failureHarald Nordgren via GitGitGadget, Oct 3, 2026
  29. 2/2 ci: point test failures and fixed known breakages at their file and lineHarald Nordgren via GitGitGadget, Oct 3, 2026
  30. Phillip WoodOct 3, 2026
  31. Phillip WoodOct 3, 2026
  32. Harald NordgrenOct 4, 2026
  33. Phillip WoodOct 5, 2026
  34. Harald NordgrenOct 5, 2026
  35. Phillip WoodOct 5, 2026
  36. Harald NordgrenOct 4, 2026
  37. 0/2 ci: link failure and leak annotations to the test scriptHarald Nordgren via GitGitGadget, Oct 6, 2026
  38. 1/2 ci: annotate leaks and stop a leak-sanitizer script at its first failureHarald Nordgren via GitGitGadget, Oct 6, 2026
  39. 2/2 ci: point test failures and fixed known breakages at their file and lineHarald Nordgren via GitGitGadget, Oct 6, 2026
  40. Phillip WoodOct 6, 2026

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.