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
$ cat > adddrop.py <<EOF
> from mercurial import registrar
> cmdtable = {}
> command = registrar.command(cmdtable)
> @command(b'debugadddrop',
> [(b'', b'drop', False, b'drop file from dirstate', b'FILE'),
> (b'', b'normal-lookup', False, b'add file to dirstate', b'FILE')],
> b'hg debugadddrop')
> def debugadddrop(ui, repo, *pats, **opts):
> '''Add or drop unnamed arguments to or from the dirstate'''
> drop = opts.get('drop')
> nl = opts.get('normal_lookup')
> if nl and drop:
> raise error.Abort('drop and normal-lookup are mutually exclusive')
> wlock = repo.wlock()
> try:
> for file in pats:
> if opts.get('normal_lookup'):
> with repo.dirstate.parentchange():
> repo.dirstate.update_file(
> file,
> p1_tracked=True,
> wc_tracked=True,
> possibly_dirty=True,
> )
> else:
> repo.dirstate._map.reset_state(file)
> repo.dirstate._dirty = True
>
> repo.dirstate.write(repo.currenttransaction())
> finally:
> wlock.release()
> EOF
$ echo "[extensions]" >> $HGRCPATH
$ echo "debugadddrop=`pwd`/adddrop.py" >> $HGRCPATH
basic test for hg debugrebuildstate
$ hg init repo
$ cd repo
$ touch foo bar
$ hg ci -Am 'add foo bar'
adding bar
adding foo
$ touch baz
$ hg add baz
$ hg rm bar
$ hg debugrebuildstate
state dump after
$ hg debugstate --no-dates | sort
n 0 -1 unset bar
n 0 -1 unset foo
$ hg debugadddrop --normal-lookup file1 file2
$ hg debugadddrop --drop bar
$ hg debugadddrop --drop
$ hg debugstate --no-dates
n 0 -1 unset file1
n 0 -1 unset file2
n 0 -1 unset foo
$ hg debugrebuildstate
status
$ hg st -A
! bar
? baz
C foo
Test debugdirstate --minimal where a file is not in parent manifest
but in the dirstate
$ touch foo bar qux
$ hg add qux
$ hg remove bar
$ sleep 1 # remove potential ambiguity in mtime
$ hg status -A
A qux
R bar
? baz
C foo
$ hg debugadddrop --normal-lookup baz
$ hg debugdirstate --no-dates
r 0 0 * bar (glob)
n 0 -1 * baz (glob)
n 644 0 * foo (glob)
a 0 -1 * qux (glob)
$ hg debugrebuilddirstate --minimal
$ hg debugdirstate --no-dates
r 0 0 * bar (glob)
n 644 0 * foo (glob)
a 0 -1 * qux (glob)
$ hg status -A
A qux
R bar
? baz
C foo
Test debugdirstate --minimal where file is in the parent manifest but not the
dirstate
$ hg manifest
bar
foo
$ sleep 1 # remove potential ambiguity in mtime
$ hg status -A
A qux
R bar
? baz
C foo
$ hg debugdirstate --no-dates
r 0 0 * bar (glob)
n 644 0 * foo (glob)
a 0 -1 * qux (glob)
$ hg debugadddrop --drop foo
$ hg debugdirstate --no-dates
r 0 0 * bar (glob)
a 0 -1 * qux (glob)
$ hg debugrebuilddirstate --minimal
$ hg debugdirstate --no-dates
r 0 0 * bar (glob)
n 0 -1 * foo (glob)
a 0 -1 * qux (glob)
$ hg status -A
A qux
R bar
? baz
C foo