Re: [PATCH v2 2/8] pack-objects: ensure tree/tag closure with '--stdin-packs=follow'

From: Jeff King <peff@peff.net>
Date: Oct 3, 2026, 01:06 UTC
Message-ID: <20261003010613.GA839051@coredump.intra.peff.net>
In-Reply-To: <asBShFkQRWJX4RQU@com-79390>

On Fri, Oct 02, 2026 at 07:55:32PM -0500, Taylor Blau wrote:

Show 7 quoted lines

> On Fri, Oct 02, 2026 at 07:13:36PM -0400, Jeff King wrote:
> > So it would have made more sense to me to comment it there. Of course
> > that is hard when there are two such places.
> >
> > I dunno.
> 
> Yeah, me either. I'm happy to change things around if you feel strongly.

I don't. If there were an easy solution I probably would. ;)

Show 7 quoted lines

> > BTW, is it safe to prepare_revision_walk() twice on the same rev_info? I
> > could believe it works, but I could also believe that there are hidden
> > corner cases, as I don't think it was ever really intended to work this
> > way.
> [...]
> Just as well, there are a couple of spots that I was able to find that
> already call `prepare_revision_walk()` more than once:

OK. That makes me feel like we're in good company, at least. If some combination turns out to be a problem, we can deal with it later.

Show 7 quoted lines

>   * In builtin/pack-objects.c::get_object_list() (with the exception of
>     '--path-walk') we call `prepare_revision_walk()` twice when
>     exploding unreachable objects as loose.
> 
>   * In reachable.c::mark_reachable_objects(), we also call the
>     `prepare_revision_walk()` function twice when given a timestamp via
>     `mark_recent`.

I have a feeling that least one of those is my fault, too. ;)

-Peff