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
# revlogutils/debug.py - utility used for revlog debuging
#
# Copyright 2005-2007 Olivia Mackall <olivia@selenic.com>
# Copyright 2022 Octobus <contact@octobus.net>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
from .. import (
node as nodemod,
)
from . import (
constants,
)
INDEX_ENTRY_DEBUG_COLUMN = []
NODE_SIZE = object()
class _column_base:
"""constains the definition of a revlog column
name: the column header,
value_func: the function called to get a value,
size: the width of the column,
verbose_only: only include the column in verbose mode.
"""
def __init__(self, name, value_func, size=None, verbose=False):
self.name = name
self.value_func = value_func
if size is not NODE_SIZE:
if size is None:
size = 8 # arbitrary default
size = max(len(name), size)
self._size = size
self.verbose_only = verbose
def get_size(self, node_size):
if self._size is NODE_SIZE:
return node_size
else:
return self._size
def debug_column(name, size=None, verbose=False):
"""decorated function is registered as a column
name: the name of the column,
size: the expected size of the column.
"""
def register(func):
entry = _column_base(
name=name,
value_func=func,
size=size,
verbose=verbose,
)
INDEX_ENTRY_DEBUG_COLUMN.append(entry)
return entry
return register
@debug_column(b"rev", size=6)
def _rev(index, rev, entry, hexfn):
return b"%d" % rev
@debug_column(b"rank", size=6, verbose=True)
def rank(index, rev, entry, hexfn):
return b"%d" % entry[constants.ENTRY_RANK]
@debug_column(b"linkrev", size=6)
def _linkrev(index, rev, entry, hexfn):
return b"%d" % entry[constants.ENTRY_LINK_REV]
@debug_column(b"nodeid", size=NODE_SIZE)
def _nodeid(index, rev, entry, hexfn):
return hexfn(entry[constants.ENTRY_NODE_ID])
@debug_column(b"p1-rev", size=6, verbose=True)
def _p1_rev(index, rev, entry, hexfn):
return b"%d" % entry[constants.ENTRY_PARENT_1]
@debug_column(b"p1-nodeid", size=NODE_SIZE)
def _p1_node(index, rev, entry, hexfn):
parent = entry[constants.ENTRY_PARENT_1]
p_entry = index[parent]
return hexfn(p_entry[constants.ENTRY_NODE_ID])
@debug_column(b"p2-rev", size=6, verbose=True)
def _p2_rev(index, rev, entry, hexfn):
return b"%d" % entry[constants.ENTRY_PARENT_2]
@debug_column(b"p2-nodeid", size=NODE_SIZE)
def _p2_node(index, rev, entry, hexfn):
parent = entry[constants.ENTRY_PARENT_2]
p_entry = index[parent]
return hexfn(p_entry[constants.ENTRY_NODE_ID])
@debug_column(b"full-size", size=20, verbose=True)
def full_size(index, rev, entry, hexfn):
return b"%d" % entry[constants.ENTRY_DATA_UNCOMPRESSED_LENGTH]
@debug_column(b"delta-base", size=6, verbose=True)
def delta_base(index, rev, entry, hexfn):
return b"%d" % entry[constants.ENTRY_DELTA_BASE]
@debug_column(b"flags", size=2, verbose=True)
def flags(index, rev, entry, hexfn):
field = entry[constants.ENTRY_DATA_OFFSET]
field &= 0xFFFF
return b"%d" % field
@debug_column(b"comp-mode", size=4, verbose=True)
def compression_mode(index, rev, entry, hexfn):
return b"%d" % entry[constants.ENTRY_DATA_COMPRESSION_MODE]
@debug_column(b"data-offset", size=20, verbose=True)
def data_offset(index, rev, entry, hexfn):
field = entry[constants.ENTRY_DATA_OFFSET]
field >>= 16
return b"%d" % field
@debug_column(b"chunk-size", size=10, verbose=True)
def data_chunk_size(index, rev, entry, hexfn):
return b"%d" % entry[constants.ENTRY_DATA_COMPRESSED_LENGTH]
@debug_column(b"sd-comp-mode", size=7, verbose=True)
def sidedata_compression_mode(index, rev, entry, hexfn):
compression = entry[constants.ENTRY_SIDEDATA_COMPRESSION_MODE]
if compression == constants.COMP_MODE_PLAIN:
return b"plain"
elif compression == constants.COMP_MODE_DEFAULT:
return b"default"
elif compression == constants.COMP_MODE_INLINE:
return b"inline"
else:
return b"%d" % compression
@debug_column(b"sidedata-offset", size=20, verbose=True)
def sidedata_offset(index, rev, entry, hexfn):
return b"%d" % entry[constants.ENTRY_SIDEDATA_OFFSET]
@debug_column(b"sd-chunk-size", size=10, verbose=True)
def sidedata_chunk_size(index, rev, entry, hexfn):
return b"%d" % entry[constants.ENTRY_SIDEDATA_COMPRESSED_LENGTH]
def debug_index(
ui,
repo,
formatter,
revlog,
full_node,
):
"""display index data for a revlog"""
if full_node:
hexfn = nodemod.hex
else:
hexfn = nodemod.short
idlen = 12
for i in revlog:
idlen = len(hexfn(revlog.node(i)))
break
fm = formatter
header_pieces = []
for column in INDEX_ENTRY_DEBUG_COLUMN:
if column.verbose_only and not ui.verbose:
continue
size = column.get_size(idlen)
name = column.name
header_pieces.append(name.rjust(size))
fm.plain(b' '.join(header_pieces) + b'\n')
index = revlog.index
for rev in revlog:
fm.startitem()
entry = index[rev]
first = True
for column in INDEX_ENTRY_DEBUG_COLUMN:
if column.verbose_only and not ui.verbose:
continue
if not first:
fm.plain(b' ')
first = False
size = column.get_size(idlen)
value = column.value_func(index, rev, entry, hexfn)
display = b"%%%ds" % size
fm.write(column.name, display, value)
fm.plain(b'\n')
fm.end()