Jeff King [off-list ref] 于2021年9月5日周日 下午8:49写道：

On Sun, Sep 05, 2021 at 04:19:53PM +0800, ZheNing Hu wrote:

quoted

quoted

In this version there are 2 patches, tested against 'git for-each-ref
--format="%(objectname) %(refname)"' on a fully packed repo with 500k
refs:

Regarding this 500k refs, is there any way I can reproduce it?

Try this in a clone of linux.git (or any other repo):

  git rev-list HEAD |
  head -500000 |
  perl -lne 'print "create refs/foo/$. $_"' |
  git update-ref --stdin

  git pack-refs --all --prune

Sorry, It seems that the above command is difficult to complete on my
machine (it took more than ten minutes). It may be stuck on git update-ref.
So I tried to reproduce it in a repo which containing 76K refs:

Benchmark #1: jk-for-each-ref-speedup~2: git for-each-ref
--format='%(refname) %(objectname)'
 Time (mean ± σ):     108.0 ms ±   1.9 ms    [User: 55.2 ms, System: 52.1 ms]
 Range (min … max):   105.7 ms … 112.4 ms    26 runs

Benchmark #2: jk-for-each-ref-speedup~1: git for-each-ref
--format='%(refname) %(objectname)'
 Time (mean ± σ):      88.2 ms ±   1.7 ms    [User: 44.8 ms, System: 43.1 ms]
 Range (min … max):    85.8 ms …  93.2 ms    32 runs

Benchmark #3:jk-for-each-ref-speedup: git for-each-ref
--format='%(refname) %(objectname)'
  Time (mean ± σ):      69.0 ms ±   2.0 ms    [User: 22.7 ms, System: 46.1 ms]
 Range (min … max):    66.2 ms …  74.1 ms    41 runs

For %(refname) and %(objectname), this performance optimization is
indeed amazing.

Though I actually think for these tests that it is not important that
each ref point to a unique commit (we are not opening up the objects at
all, and just treating the oids as strings).

-Peff

Thanks.
--
ZheNing Hu