This replaces the topmost commit from the earlier series that tried
to optimize N*(N-1) loop with N traversals.  With this, running the
following script in the kernel repository:

-- >8 --

git rev-list \
    --max-parents=2 --min-parents=2 --parents v3.5..v3.6-rc2 >RL

        while read result parent1 parent2
                $GIT merge-base $parent1 $parent2
        done <RL

GIT="rungit master" time sh -c "$cmd" >:stock
GIT=../git.git/git time sh -c "$cmd" >:optim
cmp :stock :optim
-- 8< --

shows a slight but measurable boost in the performance of
merge-base itself.

42.45user 3.79system 0:46.62elapsed 99%CPU (0avgtext+0avgdata 
0inputs+24outputs (0major+2708085minor)pagefaults 0swaps
39.99user 3.42system 0:43.27elapsed 100%CPU (0avgtext+0avgdata 
0inputs+24outputs (0major+2179604minor)pagefaults 0swaps

Junio C Hamano (4):
  merge_bases_many(): split out the logic to paint history
  merge-base: "--is-ancestor A B"
  in_merge_bases(): use paint_down_to_common()
  get_merge_bases_many(): walk from many tips in parallel

 builtin/merge-base.c |  21 ++++++++
 commit.c             | 143 +++++++++++++++++++++++++++++++++------------------
 2 files changed, 114 insertions(+), 50 deletions(-)


To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to
More majordomo info at

Reply via email to