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
from ..i18n import _
from .. import error
def get_checker(ui, revlog_name=b'changelog'):
"""Get a function that checks file handle position is as expected.
This is used to ensure that files haven't been modified outside of our
knowledge (such as on a networked filesystem, if `hg debuglocks` was used,
or writes to .hg that ignored locks happened).
Due to revlogs supporting a concept of buffered, delayed, or diverted
writes, we're allowing the files to be shorter than expected (the data may
not have been written yet), but they can't be longer.
Please note that this check is not perfect; it can't detect all cases (there
may be false-negatives/false-OKs), but it should never claim there's an
issue when there isn't (false-positives/false-failures).
"""
vpos = ui.config(b'debug', b'revlog.verifyposition.' + revlog_name)
# Avoid any `fh.tell` cost if this isn't enabled.
if not vpos or vpos not in [b'log', b'warn', b'fail']:
return None
def _checker(fh, fn, expected):
if fh.tell() <= expected:
return
msg = _(b'%s: file cursor at position %d, expected %d')
# Always log if we're going to warn or fail.
ui.log(b'debug', msg + b'\n', fn, fh.tell(), expected)
if vpos == b'warn':
ui.warn((msg + b'\n') % (fn, fh.tell(), expected))
elif vpos == b'fail':
raise error.RevlogError(msg % (fn, fh.tell(), expected))
return _checker