[0/2] Fix built-in rebase perf regression

Message

Bruce Perry via GitGitGadget Nov. 9, 2018, 9:34 a.m. UTC

In our tests with large repositories, we noticed a serious regression of the
performance of git rebase when using the built-in vs the shell script
version. It boils down to an incorrect conversion of a git checkout -q:
instead of using a twoway_merge as git checkout does, we used a oneway_merge 
as git reset does. The latter, however, calls lstat() on all files listed in
the index, while the former essentially looks only at the files that are
different between the given two revisions.

Let's reinstate the original behavior by introducing a flag to the 
reset_head() function to indicate whether we want to emulate reset --hard 
(in which case we use the oneway_merge, otherwise we use twoway_merge).

Johannes Schindelin (2):
  rebase: consolidate clean-up code before leaving reset_head()
  built-in rebase: reinstate `checkout -q` behavior where appropriate

 builtin/rebase.c | 60 ++++++++++++++++++++++++++----------------------
 1 file changed, 33 insertions(+), 27 deletions(-)


base-commit: 8858448bb49332d353febc078ce4a3abcc962efe
Published-As: https://github.com/gitgitgadget/git/releases/tags/pr-72%2Fdscho%2Fbuiltin-rebase-perf-regression-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-72/dscho/builtin-rebase-perf-regression-v1
Pull-Request: https://github.com/gitgitgadget/git/pull/72