Re: [PATCH v5 11/12] Fix error-prone fill_directory() API; make it only return matches
- From
Junio C Hamano <gitster@pobox.com>
- Date
- Jul 20, 2020, 18:58 UTC
- Message-ID
- <xmqq365m9fom.fsf@gitster.c.googlers.com>
- In-Reply-To
- <CABPp-BH6xYb8a5hkV_68vbXhH4kApCxw0WG9oveXOk5zzU==1A@mail.gmail.com>
Elijah Newren <newren@gmail.com> writes:
Show 16 quoted lines
>> Looking at the commit in question (95c11ecc73), there must have been >> some reason that it injected the pathspec check between the >> "path_excluded" and the "path_untracked" cases. The diff below >> basically undoes that split, so I have a feeling I'm missing something. > > Awesome, thanks Andreas for the bisected report and Martin for finding > and fixing the bug. As for the reason that the old patch injected the > pathspec check between the path_excluded and the path_untracked cases, > that appears to me to just be "I'm good at making boneheaded > mistakes". Your changes here are the right fix. As a separate > optimization, we could maybe make simplify_away() a bit more involved > and have it exclude a few more paths so that fewer make it to this > final check, but that's just optimization work that is separate from > your fix here. > > Reviewed-by: Elijah Newren <newren@gmail.com>
This is in 2.27-rc0, so it is not ultra-urgent to fix it at the tip of the current development track, but let's make sure we have a fix with a proper log messasge in a mergeable state (meaning "have already been cooked in 'next' for a week or two) early in the next cycle.
Thank you, Andreas, Martin and Elijah, as always.