bisect: avoid copying ancestor list for non-merge commits
During a bisection, hg needs to compute a list of all ancestors for every
candidate commit. This is accomplished via a bottom-up traversal of the set of
candidates, during which each revision's ancestor list is populated using the
ancestor list of its parent(s). Previously, this involved copying the entire
list, which could be very long in if the bisection range was large.
To help improve this, we can observe that each candidate commit is visited
exactly once, at which point its ancestor list is copied into its children's
lists and then dropped. In the case of non-merge commits, a commit's ancestor
list consists exactly of its parent's list plus itself. This means that we can
trivially reuse the parent's existing list for one of its non-merge children,
which avoids copying entirely if that commit is the parent's only child. This
makes bisections over linear ranges of commits much faster.
During some informal testing in the large publicly-available `mozilla-central`
repository, this noticeably sped up bisections over large ranges of history:
Setup:
$ cd mozilla-central
$ hg bisect --reset
$ hg bisect --good 0
$ hg log -r tip -T '{rev}\n'
628417
Test:
$ time hg bisect --bad tip --noupdate
Before:
real 3m35.927s
user 3m35.553s
sys 0m0.319s
After:
real 1m41.142s
user 1m40.810s
sys 0m0.285s
Testing recorded "modified" files for merge commit
==================================================
This file shows what hg says are "modified" files for a merge commit
(hg log -T {files}), somewhat exhaustively.
This file test multiple corner case.
For merges that involve files contents changing, check test-merge-combination-file-content.t
For merges that involve executable bit changing, check test-merge-combination-exec-bytes.t
Case with multiple or zero merge ancestors, copies/renames, and identical file contents
with different filelog revisions are not currently covered.
$ . $TESTDIR/testlib/merge-combination-util.sh
Files modified or cleanly merged, with no greatest common ancestors:
$ hg init repo; cd repo
$ touch a0 b0; hg commit -qAm 0
$ hg up -qr null; touch a1 b1; hg commit -qAm 1
$ hg merge -qr 0; rm b*; hg commit -qAm 2
$ hg log -r . -T '{files}\n'
b0 b1
$ cd ../
$ rm -rf repo
A few cases of criss-cross merges involving deletions (listing all
such merges is probably too much). Both gcas contain $files, so we
expect the final merge to behave like a merge with a single gca
containing $files.
$ hg init repo; cd repo
$ files="c1 u1 c2 u2"
$ touch $files; hg commit -qAm '0 root'
$ for f in $files; do echo f > $f; done; hg commit -qAm '1 gca1'
$ hg up -qr0; hg revert -qr 1 --all; hg commit -qAm '2 gca2'
$ hg up -qr 1; hg merge -qr 2; rm *1; hg commit -qAm '3 p1'
$ hg up -qr 2; hg merge -qr 1; rm *2; hg commit -qAm '4 p2'
$ hg merge -qr 3; echo f > u1; echo f > u2; rm -f c1 c2
$ hg commit -qAm '5 merge with two gcas'
$ hg log -r . -T '{files}\n' # expecting u1 u2
$ cd ../
$ rm -rf repo