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.
% init
% no working directory
% setup
0 files updated, 0 files merged, 2 files removed, 0 files unresolved
% hg children at revision 3 (tip)
0 files updated, 0 files merged, 1 files removed, 0 files unresolved
% hg children at nullrev (should be 0 and 3)
2 files updated, 0 files merged, 0 files removed, 0 files unresolved
% hg children at revision 1 (should be 2)
changeset: 2:8f5eea5023c2
user: test
date: Thu Jan 01 00:00:02 1970 +0000
summary: 2
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
% hg children at revision 2 (other head)
% hg children -r null
changeset: 0:4df8521a7374
user: test
date: Thu Jan 01 00:00:00 1970 +0000
summary: 0
changeset: 3:e2962852269d
tag: tip
parent: -1:000000000000
user: test
date: Thu Jan 01 00:00:03 1970 +0000
summary: 3
% hg children -r 0
changeset: 1:708c093edef0
user: test
date: Thu Jan 01 00:00:01 1970 +0000
summary: 1
% hg children -r 1
changeset: 2:8f5eea5023c2
user: test
date: Thu Jan 01 00:00:02 1970 +0000
summary: 2
% hg children -r 2
% hg children -r 3
% hg children -r 0 file0 (should be 2)
changeset: 2:8f5eea5023c2
user: test
date: Thu Jan 01 00:00:02 1970 +0000
summary: 2
% hg children -r 1 file0 (should be 2)
changeset: 2:8f5eea5023c2
user: test
date: Thu Jan 01 00:00:02 1970 +0000
summary: 2
1 files updated, 0 files merged, 1 files removed, 0 files unresolved
% hg children file0 at revision 0 (should be 2)
changeset: 2:8f5eea5023c2
user: test
date: Thu Jan 01 00:00:02 1970 +0000
summary: 2