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

Re: git-diff in a worktree is an order of magnitude slower?

From
Jeff King <peff@peff.net>
Date
Jun 21, 2026, 22:00 UTC
Message-ID
<20260621220046.GE2297179@coredump.intra.peff.net>
In-Reply-To
<xmqqechz60ah.fsf@gitster.g>
On Sun, Jun 21, 2026 at 02:39:18PM -0700, Junio C Hamano wrote:
Show 26 quoted lines
> Jeff King <peff@peff.net> writes:
> 
> > BTW, I don't think diffcore actually has the information it would need
> > to do so. The racy stuff is handled under the hood in ie_match_stat(),
> > which returns only a set of "changed" flags. So the caller cannot tell
> > the difference between the two cases:
> >
> >   1. We checked ce_match_stat_basic() which said "no change", and then
> >      is_racy_timestamp() was false, so that was good enough.
> >
> >   2. is_racy_timestamp() is true, so we further did a content check,
> >      found nothing, and returned the same "no change"
> >
> > Obviously we could pass back another flag, but that would disrupt the
> > other callers. Hmm. It looks like we could pass in a flag to say "assume
> > racy entries are modified". And then they come back to the diff code,
> > diffcore_skip_stat_unmatch() sees they're not real diffs and suppresses
> > them, but we _do_ count them as stat-dirty.
> 
> Yeah.  Because ie_match_stat() does have access to istate, we could
> add a new member to istate, next to "updated_workdir" and friends,
> and smudge the bit when the is_racy_timestamp() goes to the
> compare-data codepath and finds that we are better off auto
> refreshing.  Then "were we told to do skip-stat-unmatch and actually
> found some that is worth refreshing?" code can be taught to pay
> attention to that bit as well.

Yeah, that sounds fairly clean. Though if using nanoseconds works out and makes racy entries extremely unlikely, that is better still. :)

> This is a tangent, but why do we call refresh_index_quietly() in the
> central code path in cmd_diff() in the first place, I have to
> wonder?  It should not matter when we are comparing two tree objects
> (or two commits), at least.  It of course is not hurting, though.

It seems like it could probably just go into builtin_diff_files(), but are there other paths that might hit stat-unmatch entries? Maybe the builtin_diff_b_f() path?

It probably should also support --no-optional-locks, which is currently only respected by git-status. I don't think it matters that much in practice because the point is reducing conflict with commands running frequently in the background, and people don't tend to do that with git-diff.

Back when we added --no-optional-locks, the idea was that people could apply it in more spots if they ran into them in practice. So I guess nobody has with git-diff.

-Peff
Previous: Junio C HamanoNext: D. Ben Knoble
Message 18 of 19 in “git-diff in a worktree is an order of magnitude slower?”
  1. D. Ben KnobleJun 8, 2026
  2. Jeff KingJun 9, 2026
  3. D. Ben KnobleJun 9, 2026
  4. Jeff KingJun 11, 2026
  5. Junio C HamanoJun 11, 2026
  6. brian m. carlsonJun 11, 2026
  7. D. Ben KnobleJun 20, 2026
  8. Junio C HamanoJun 21, 2026
  9. Junio C HamanoJun 21, 2026
  10. Jeff KingJun 21, 2026
  11. Jeff KingJun 21, 2026
  12. Junio C HamanoJun 21, 2026
  13. Jeff KingJun 21, 2026
  14. Junio C HamanoJun 21, 2026
  15. Junio C HamanoJun 22, 2026
  16. Jeff KingJun 28, 2026
  17. Junio C HamanoJun 21, 2026
  18. Jeff KingJun 21, 2026
  19. D. Ben KnobleJul 3, 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.