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

Re: [PATCH 00/20] Separate `ref_cache` into a separate module

From
Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Date
Mar 20, 2017, 18:24 UTC
Message-ID
<CACBZZX7Kqo=w-c2LOH9a4xf6WpnhE5tsvNHa1O2-4dvYTKZxEA@mail.gmail.com>
In-Reply-To
<cover.1490026594.git.mhagger@alum.mit.edu>
On Mon, Mar 20, 2017 at 5:33 PM, Michael Haggerty <mhagger@alum.mit.edu> wrote:
Show 5 quoted lines
> I've completed a draft of an epic 48-patch series implementing all of
> the above points on my GitHub fork [1] as branch
> `wip/mmap-packed-refs`. It dramatically speeds up performance and
> reduces memory usage for some tasks in repositories with very many
> packed references.

Just curious, what sort of performance numbers on mainly what sort of operations? I'd expect the fixed cost of e.g. `git log -1 >/dev/null` to go down, which can take as long as a second on some repos with a huge amount of refs, but anything else?

Previous: Jeff KingNext: Jeff King
Message 35 of 38 in “Separate `ref_cache` into a separate module”
  1. 00/20 Separate `ref_cache` into a separate moduleMichael Haggerty, Mar 20, 2017
  2. 01/20 get_ref_dir(): don't call read_loose_refs() for "refs/bisect"Michael Haggerty, Mar 20, 2017
  3. 10/20 ref-cache: introduce a new type, ref_cacheMichael Haggerty, Mar 20, 2017
  4. 13/20 refs: handle "refs/bisect/" in `loose_fill_ref_dir()`Michael Haggerty, Mar 20, 2017
  5. 11/20 refs: record the ref_store in ref_cache, not ref_dirMichael Haggerty, Mar 20, 2017
  6. Jeff KingMar 20, 2017
  7. Michael HaggertyMar 20, 2017
  8. 08/20 Rename `remove_entry()` to `remove_entry_from_dir()`Michael Haggerty, Mar 20, 2017
  9. 18/20 commit_packed_refs(): use reference iterationMichael Haggerty, Mar 20, 2017
  10. Jeff KingMar 20, 2017
  11. Michael HaggertyMar 22, 2017
  12. Jeff KingMar 22, 2017
  13. 20/20 do_for_each_entry_in_dir(): delete functionMichael Haggerty, Mar 20, 2017
  14. 16/20 get_loose_ref_cache(): new functionMichael Haggerty, Mar 20, 2017
  15. 15/20 get_loose_ref_dir(): function renamed from get_loose_refs()Michael Haggerty, Mar 20, 2017
  16. 19/20 files_pack_refs(): use reference iterationMichael Haggerty, Mar 20, 2017
  17. 14/20 do_for_each_entry_in_dir(): eliminate `offset` argumentMichael Haggerty, Mar 20, 2017
  18. 03/20 refs_ref_iterator_begin(): new functionMichael Haggerty, Mar 20, 2017
  19. 17/20 cache_ref_iterator_begin(): make function smarterMichael Haggerty, Mar 20, 2017
  20. 07/20 Rename `find_ref()` to `find_ref_entry()`Michael Haggerty, Mar 20, 2017
  21. 12/20 ref-cache: use a callback function to fill the cacheMichael Haggerty, Mar 20, 2017
  22. 09/20 refs: split `ref_cache` code into separate filesMichael Haggerty, Mar 20, 2017
  23. Jeff KingMar 20, 2017
  24. Junio C HamanoMar 20, 2017
  25. Stefan BellerMar 20, 2017
  26. Junio C HamanoMar 20, 2017
  27. 04/20 refs_verify_refname_available(): implement once for all backendsMichael Haggerty, Mar 20, 2017
  28. Jeff KingMar 20, 2017
  29. Michael HaggertyMar 20, 2017
  30. 02/20 refs_read_raw_ref(): new functionMichael Haggerty, Mar 20, 2017
  31. 06/20 Rename `add_ref()` to `add_ref_entry()`Michael Haggerty, Mar 20, 2017
  32. 05/20 refs_verify_refname_available(): use function in more placesMichael Haggerty, Mar 20, 2017
  33. Junio C HamanoMar 20, 2017
  34. Jeff KingMar 20, 2017
  35. Ævar Arnfjörð BjarmasonMar 20, 2017
  36. Jeff KingMar 20, 2017
  37. Junio C HamanoMar 20, 2017
  38. Michael HaggertyMar 20, 2017

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.