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

Re: [PATCH 1/1] commit-graph.c: avoid unnecessary tag dereference when merging

From
Junio C Hamano <gitster@pobox.com>
Date
Mar 21, 2020, 18:50 UTC
Message-ID
<xmqqfte1a6ew.fsf@gitster.c.googlers.com>
In-Reply-To
<20200321070333.GB1441446@coredump.intra.peff.net>
Jeff King <peff@peff.net> writes:
Show 5 quoted lines
> So we weren't actually reading all of the commits even under the old
> code. We were just going into deref_tag(), seeing that the object is
> already parsed, and then quickly returning when we see that we already
> have an OBJ_COMMIT. I suspect most of your timing differences are mostly
> noise.
Noise that contributes largely to ~7% fluctuation?  That sounds big.
Show 9 quoted lines
> I guess --input=stdin-commits is a good way to simulate that. Try this
> (assuming there's already a split-graph file with all of the commits in
> it):
>
>   git rev-parse HEAD >input
>   time git commit-graph write --input=stdin-commits --split=merge-all <input
> ...
> A more realistic case would probably be feeding a new small pack to
> --input=stdin-packs.
Makes sense.
Show 12 quoted lines
> At any rate, I think there is a demonstrable speedup there. But
> moreover, I'm pretty sure this existing code is not doing what it
> expects:
>
>   /* only add commits if they still exist in the repo */
>   result = lookup_commit_reference_gently(ctx->r, &oid, 1);
>
> That won't look at the object database at all if the commit is already
> marked as parsed. And that parsing might have come from the commit graph
> itself, as our earlier attempts showed. So switching to a real
> has_object_file() call is an important _correctness_ fix, even leaving
> aside the speed improvements.

True. has_object_file() asks the question without even lookup-replace flag set, so it would try to see if the object is truly there (modulo the "pretended" ones are declared to exist via the cached-object mechanism).

Do we need to worry about INFO_QUICK and SKIP_FETCH_OBJECT in this codepath, by the way?

Previous: Jeff KingNext: Derrick Stolee
Message 15 of 31 in “commit-graph: avoid unnecessary tag deference when merging”
  1. 0/1 commit-graph: avoid unnecessary tag deference when mergingTaylor Blau, Mar 21, 2020
  2. 1/1 commit-graph.c: avoid unnecessary tag dereference when mergingTaylor Blau, Mar 21, 2020
  3. Jeff KingMar 21, 2020
  4. Taylor BlauMar 21, 2020
  5. Taylor BlauMar 21, 2020
  6. Jeff KingMar 21, 2020
  7. Taylor BlauMar 21, 2020
  8. Jeff KingMar 22, 2020
  9. SZEDER GáborMar 22, 2020
  10. looking up object types quickly, was Re: [PATCH 1/1] commit-graph.c: avoid unnecessary tag dereference when mergingJeff King, Mar 22, 2020
  11. Jeff KingMar 22, 2020
  12. Taylor BlauMar 23, 2020
  13. Taylor BlauMar 22, 2020
  14. Jeff KingMar 24, 2020
  15. Junio C HamanoMar 21, 2020
  16. Derrick StoleeMar 22, 2020
  17. Taylor BlauMar 22, 2020
  18. Derrick StoleeMar 22, 2020
  19. Jeff KingMar 22, 2020
  20. Jeff KingMar 22, 2020
  21. Taylor BlauMar 22, 2020
  22. Jeff KingMar 24, 2020
  23. Taylor BlauMar 24, 2020
  24. Jeff KingMar 27, 2020
  25. Taylor BlauMar 27, 2020
  26. Taylor BlauMar 22, 2020
  27. Jeff KingMar 24, 2020
  28. Junio C HamanoMar 21, 2020
  29. Junio C HamanoMar 21, 2020
  30. Jeff KingMar 21, 2020
  31. Taylor BlauMar 21, 2020

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.