Mercurial > hg
view tests/test-import-context.t @ 24560:b38bcf18993c
dirstate.walk: don't keep track of normalized files in parallel
Rev 2bb13f2b778c changed the semantics of the work list to store (normalized,
non-normalized) pairs. All the tuple creation and destruction hurts perf: on a
large repo on OS X, 'hg status' went from 3.62 seconds to 3.78.
It also is unnecessary in most cases:
- it is clearly unnecessary on case-sensitive filesystems.
- it is also unnecessary when filenames have been read off of disk rather than
being supplied by the user.
The only case where the non-normalized case is required at all is when the file
is unknown.
To eliminate most of the perf cost, keep trace of whether the directory needs
to be normalized at all with a boolean called 'alreadynormed'. Pay the cost of
directory normalization only when necessary.
For the above large repo, 'hg status' goes to 3.63 seconds.
author | Siddharth Agarwal <sid0@fb.com> |
---|---|
date | Tue, 31 Mar 2015 19:29:39 -0700 |
parents | f2719b387380 |
children | 75be14993fda |
line wrap: on
line source
Test applying context diffs $ cat > writepatterns.py <<EOF > import sys > > path = sys.argv[1] > lasteol = sys.argv[2] == '1' > patterns = sys.argv[3:] > > fp = file(path, 'wb') > for i, pattern in enumerate(patterns): > count = int(pattern[0:-1]) > char = pattern[-1] + '\n' > if not lasteol and i == len(patterns) - 1: > fp.write((char*count)[:-1]) > else: > fp.write(char*count) > fp.close() > EOF $ cat > cat.py <<EOF > import sys > sys.stdout.write(repr(file(sys.argv[1], 'rb').read()) + '\n') > EOF Initialize the test repository $ hg init repo $ cd repo $ python ../writepatterns.py a 0 5A 1B 5C 1D $ python ../writepatterns.py b 1 1A 1B $ python ../writepatterns.py c 1 5A $ python ../writepatterns.py d 1 5A 1B $ hg add adding a adding b adding c adding d $ hg ci -m addfiles Add file, missing a last end of line $ hg import --no-commit - <<EOF > *** /dev/null 2010-10-16 18:05:49.000000000 +0200 > --- b/newnoeol 2010-10-16 18:23:26.000000000 +0200 > *************** > *** 0 **** > --- 1,2 ---- > + a > + b > \ No newline at end of file > *** a/a Sat Oct 16 16:35:51 2010 > --- b/a Sat Oct 16 16:35:51 2010 > *************** > *** 3,12 **** > A > A > A > ! B > C > C > C > C > C > ! D > \ No newline at end of file > --- 3,13 ---- > A > A > A > ! E > C > C > C > C > C > ! F > ! F > > *** a/b 2010-10-16 18:40:38.000000000 +0200 > --- /dev/null 2010-10-16 18:05:49.000000000 +0200 > *************** > *** 1,2 **** > - A > - B > --- 0 ---- > *** a/c Sat Oct 16 21:34:26 2010 > --- b/c Sat Oct 16 21:34:27 2010 > *************** > *** 3,5 **** > --- 3,7 ---- > A > A > A > + B > + B > *** a/d Sat Oct 16 21:47:20 2010 > --- b/d Sat Oct 16 21:47:22 2010 > *************** > *** 2,6 **** > A > A > A > - A > - B > --- 2,4 ---- > EOF applying patch from stdin $ hg st M a M c M d A newnoeol R b What's in a $ python ../cat.py a 'A\nA\nA\nA\nA\nE\nC\nC\nC\nC\nC\nF\nF\n' $ python ../cat.py newnoeol 'a\nb' $ python ../cat.py c 'A\nA\nA\nA\nA\nB\nB\n' $ python ../cat.py d 'A\nA\nA\nA\n' $ cd ..