From: Harald Nordgren <haraldnordgren@gmail.com>
Date: Mon, 05 Oct 2026 13:59:48 GMT
Subject: Re: [PATCH v5 0/2] ci: link failure and leak annotations to the test script
Message-ID: <CAHwyqnUO0zvr+hPT2t0CG-7D9vZuWBRdj1RjC356WEuXaZ9Faw@mail.gmail.com>
In-Reply-To: <ea988ec0-ef3d-4250-a0d6-ffdf3794b9cf@gmail.com>

> > Is this enough to call this a regression? Then maybe it's not worth
> > doing this part at all.
>
> Yes, I think we should drop this patch. The first step to debugging a
> test failure is to look at the test output, so the current behavior
> where clicking on the links on the summary page takes you to the test
> output is more useful than taking you to a diff that may not even show
> the test that failed. The first patch is definitely worth keeping as it
> makes it much easier to see the LSAN output.

I played with instead showing the file name (and line when available)
as part of the annotation text, and leaving the linking as it is. I
think it could gives us the best of both worlds:

    memory leak logged in t1060 (t1060-object-corruption.sh)

and

    failed: t1060.17 partial clone of corrupted repository
(t1060-object-corruption.sh:141)

What do you think?


Harald