Re: [PATCH 0/2] Squash leaks in t0000

From: Carlo Arenas <carenas@gmail.com>
Date: Sep 18, 2021, 17:28 UTC
Message-ID: <CAPUEspjjBAr07VB7XqGFVXDcKWkkX4OUTCxUf+VJaEAX8KTAyw@mail.gmail.com>
In-Reply-To: <pull.1092.git.git.1631972978.gitgitgadget@gmail.com>

On Sat, Sep 18, 2021 at 6:49 AM Andrzej Hunt via GitGitGadget <gitgitgadget@gmail.com> wrote:

> Carlo points out that t0000 currently doesn't pass with leak-checking
> enabled in:
> https://public-inbox.org/git/CAPUEsphMUNYRACmK-nksotP1RrMn09mNGFdEHLLuNEWH4AcU7Q@mail.gmail.com/T/#m7e40220195d98aee4be7e8593d30094b88a6ee71

Did you figure out why it doesn't trigger on maint even if the code seems to be mostly the same? At least seems to trigger consistently in master, next and seen.

> Here's a series that I've sat on for a while, which adds some UNLEAK's to
> "fix" this situation - see the individual patches for a justification of why
> an UNLEAK seems appropriate.

While I see that UNLEAK in this specific case, might be an ok "fix", I have to admit that not finding a repo_clear_revisions() (or equivalent function) that could be used to clear revs seems like a problem worth fixing as well for the future.

Will reply with my WIP so we can see if it could work either as an alternative to this, or at least lay some foundations so that a long running process that needs to use a `struct revision` or some of this logic can in the future without having to deal with leaks.

Thanks and "Reviewed-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>" if needed.

Carlo