view tests/test-git-import.out @ 5339:058e93c3d07d

I have spotted the biggest bottleneck in "bdiff.c". Actually it was
pretty easy to find after I recompiled the python interpreter and
mercurial for profiling.

In "bdiff.c" function "equatelines" allocates the minimum hash table
size, which can lead to tons of collisions. I introduced an
"overcommit" factor of 16, this is, I allocate 16 times more memory
than the minimum value. Overcommiting 128 times does not improve the
performance over the 16-times case.

% new file
applying patch from stdin
% new empty file
applying patch from stdin
empty
% chmod +x
applying patch from stdin
% copy
applying patch from stdin
a
a
% rename
applying patch from stdin
copyx
empty
new
rename
% delete
applying patch from stdin
empty
new
rename
% regular diff
applying patch from stdin
% copy and modify
applying patch from stdin
a
a
b
a
a
% rename and modify
applying patch from stdin
a
a
b
c
a
% one file renamed multiple times
applying patch from stdin
9 rename2 rename3 rename3-2 / rename3 (rename2)rename3-2 (rename2)
rename3
rename3-2
a
a
b
c
a

a
a
b
c
a
% binary files and regular patch hunks
applying patch from stdin
foo
045c85ba38952325e126c70962cc0f9d9077bc67 644 binary
% many binary files
applying patch from stdin
045c85ba38952325e126c70962cc0f9d9077bc67 644 mbinary1
a874b471193996e7cb034bb301cac7bdaf3e3f46 644 mbinary2
% filenames with spaces
applying patch from stdin
foo