view hgext/extdiff.py @ 39506:b66ea3fc3a86

sparse-revlog: set max delta chain length to on thousand The new snapshot system used in the sparse-revlog case gave us some small size benefit so far. However its most important property is to gracefully handle harder limit on delta chainlength. Long delta chain has a very detrimental impact on read (and write) performance in revlog. Being able to shorter them provide a great boost. However, shorting delta used to result significantly lower compression ratio. The intermediate snapshots effectively suppress most of this effect (even all in some case). # Effect on the test repository The repository we use for test is not "realistic" but can still show this in action using an unreasonably low chain limit. Limiting the chain length show a sizeable increase but stay under control: +6% for limit=15; +15% for limit=10. Without the snapshot system the increase is significantly bigger: +45% for limit=15; +80% for limit=10. Even slightly larger than without delta chain limit, the resulting size is still smaller than before we started doing snapshots. Here is a table for comparison. *Since the repository is not branchy, the initial sparse-revlog version does not bring much benefit compare to the non-sparse one): chain length limit | none | limit=15 | limit=10 | without sparse-revlog | 62 818 987 | 112 664 615 | 131 222 574 | without snapshot | 74 365 490 | 108 211 410 | 133 857 764 | with snapshot | 59 230 936 | 63 002 924 | 68 415 329 | # Effect On Real Life Repositories The series provides significant benefits on all kind of repositories. Using `hg debugupgraderepo -o redeltaparent --run`, we recomputed delta chain for various repositories with different settings: - delta chain length: unlimited or 1000 limit - sparse-revlog: enabled or disabled - this series: applied or not applied We can observe multiple types of effect: - On very branchy repositories: * The delta chain limit as low impact on the repo size. * Intermediate snapshot greatly reduces manifest size: - pypy: -80% - netbeans: -95% * The delta chain limit is effective, without a size impact: - netbeans average: 613 -> 282 - private #1 average: 1 068 -> 307 - On more linear repository: * Intermediate snapshot limit the impact of delta chain limit: - mozilla: without the series: +360% with the series: +25% * The delta chain limit provides large improvement: - mozilla's average chain length: unlimited: 15 338 limited: 469 * Despite the chain length limit, the manifest size is reduced: - mercurial: -25% - mozilla: -30% It is clear that the use of chains of intermediate snapshots provide large benefits both in storage size and delta chains quality. We should now switch our effort toward making sure the write performance are acceptable. Then, `sparse-revlog` will be a suitable format for all new repository. # Raw Statistic * no-sparse: general delta repository not using sparse-revlog * no-snapshot: sparse-revlog repository not using this series * snapshot: sparse-revlog repository using this series mercurial Manifest Size: limit | none | 1000 ------------|-------------|------------ no-sparse | 8 021 373 | 8 199 366 no-snapshot | 8 103 561 | 8 259 719 snapshot | 6 137 116 | 6 126 433 Manifest Chain length data limit || none || 1000 || value || average | max || average | max || ------------||---------|---------||---------|---------|| no-sparse || 307 | 1456 || 279 | 1000 || no-snapshot || 312 | 1456 || 283 | 1000 || snapshot || 248 | 1208 || 241 | 1000 || Full Store Size limit | none | 1000 ------------|-------------|------------ no-sparse | 51 013 198 | 51 201 574 no-snapshot | 50 930 795 | 51 141 006 snapshot | 48 072 037 | 48 093 572 pypy Manifest Size: limit | none | 1000 ------------|-------------|------------ no-sparse | 193 987 784 | 193 987 784 no-snapshot | 163 171 745 | 163 312 229 snapshot | 34 605 900 | 34 600 750 Manifest Chain length data limit || none || 1000 || value || average | max || average | max || ------------||---------|---------||---------|---------|| no-sparse || 101 | 692 || 101 | 692 || no-snapshot || 151 | 1307 || 148 | 1000 || snapshot || 128 | 1309 || 125 | 1000 || Full Store Size limit | none | 1000 ------------|-------------|------------ no-sparse | 495 931 473 | 495 931 473 no-snapshot | 465 441 017 | 465 581 501 snapshot | 355 467 301 | 355 472 451 Mozilla Manifest Size: limit | none | 1000 ------------|----------------|--------------- no-sparse | 416 757 148 | 1 869 009 668 no-snapshot | 401 592 370 | 1 843 493 795 snapshot | 224 359 521 | 284 615 500 Manifest Chain length data limit || none || 1000 || value || average | max || average | max || ------------||---------|---------||---------|---------|| no-sparse || 15 333 | 58 980 || 468 | 1 000 || no-snapshot || 15 336 | 58 980 || 469 | 1 000 || snapshot || 15 338 | 58 983 || 469 | 1 000 || Full Store Size limit | none | 1000 ------------|----------------|--------------- no-sparse | 2 712 477 887 | 4 164 995 451 no-snapshot | 2 698 887 835 | 4 141 054 304 snapshot | 2 518 130 385 | 2 578 587 596 Netbeans Manifest Size: limit | none | 1000 ------------|----------------|--------------- no-sparse | 4 766 794 101 | 4 870 642 687 no-snapshot | 4 334 806 082 | 4 428 681 309 snapshot | 232 659 666 | 240 330 665 Manifest Chain length data limit || none || 1000 || value || average | max || average | max || ------------||---------|---------||---------|---------|| no-sparse || 597 | 6802 || 254 | 1 000 || no-snapshot || 648 | 6 802 || 305 | 1 000 || snapshot || 613 | 6 804 || 282 | 1 000 || Full Store Size limit | none | 1000 ------------|----------------|--------------- no-sparse | 5 807 347 998 | 5 911 196 584 no-snapshot | 5 375 398 602 | 5 469 273 829 snapshot | 1 282 519 928 | 1 290 190 927 Private repo #1 Manifest Size: limit | none | 1000 ------------|-----------------|--------------- no-sparse | 41 389 010 840 | 41 398 162 091 no-snapshot | 9 737 319 435 | 10 223 773 150 snapshot | 744 215 807 | 747 961 822 Manifest Chain length data limit || none || 1000 || value || average | max || average | max || ------------||---------|---------||---------|---------|| no-sparse || 245 | 8 885 || 81 | 1 000 || no-snapshot || 1 225 | 8 885 || 336 | 1 000 || snapshot || 1 068 | 7 909 || 307 | 1 000 || Full Store Size limit | none | 1000 ------------|----------------|--------------- no-sparse | 49 646 065 126 | 49 655 216 377 no-snapshot | 17 924 862 856 | 18 411 316 571 snapshot | 9 009 024 710 | 9 012 770 725 Private repo #2 We currently have less data available for that repository. * Before is a sparse-revlog repository without this series * After is a sparse-revlog repository with this series + 1000 chain limit Manifest Size: Before: 1 531 485 040 bytes After: 1 091 422 451 bytes Manifest Chain: Before: 2 218 avg; 6 575 Max After: 442 avg; 1 000 Max Full Store Size Before: 15 203 955 615 after: 8 207 180 693
author Boris Feld <boris.feld@octobus.net>
date Fri, 07 Sep 2018 11:18:45 -0400
parents 2ce60954b1b7
children c303d65d2e34
line wrap: on
line source

# extdiff.py - external diff program support for mercurial
#
# Copyright 2006 Vadim Gelfer <vadim.gelfer@gmail.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.

'''command to allow external programs to compare revisions

The extdiff Mercurial extension allows you to use external programs
to compare revisions, or revision with working directory. The external
diff programs are called with a configurable set of options and two
non-option arguments: paths to directories containing snapshots of
files to compare.

If there is more than one file being compared and the "child" revision
is the working directory, any modifications made in the external diff
program will be copied back to the working directory from the temporary
directory.

The extdiff extension also allows you to configure new diff commands, so
you do not need to type :hg:`extdiff -p kdiff3` always. ::

  [extdiff]
  # add new command that runs GNU diff(1) in 'context diff' mode
  cdiff = gdiff -Nprc5
  ## or the old way:
  #cmd.cdiff = gdiff
  #opts.cdiff = -Nprc5

  # add new command called meld, runs meld (no need to name twice).  If
  # the meld executable is not available, the meld tool in [merge-tools]
  # will be used, if available
  meld =

  # add new command called vimdiff, runs gvimdiff with DirDiff plugin
  # (see http://www.vim.org/scripts/script.php?script_id=102) Non
  # English user, be sure to put "let g:DirDiffDynamicDiffText = 1" in
  # your .vimrc
  vimdiff = gvim -f "+next" \\
            "+execute 'DirDiff' fnameescape(argv(0)) fnameescape(argv(1))"

Tool arguments can include variables that are expanded at runtime::

  $parent1, $plabel1 - filename, descriptive label of first parent
  $child,   $clabel  - filename, descriptive label of child revision
  $parent2, $plabel2 - filename, descriptive label of second parent
  $root              - repository root
  $parent is an alias for $parent1.

The extdiff extension will look in your [diff-tools] and [merge-tools]
sections for diff tool arguments, when none are specified in [extdiff].

::

  [extdiff]
  kdiff3 =

  [diff-tools]
  kdiff3.diffargs=--L1 '$plabel1' --L2 '$clabel' $parent $child

You can use -I/-X and list of file or directory names like normal
:hg:`diff` command. The extdiff extension makes snapshots of only
needed files, so running the external diff program will actually be
pretty fast (at least faster than having to compare the entire tree).
'''

from __future__ import absolute_import

import os
import re
import shutil
import stat

from mercurial.i18n import _
from mercurial.node import (
    nullid,
    short,
)
from mercurial import (
    archival,
    cmdutil,
    error,
    filemerge,
    formatter,
    pycompat,
    registrar,
    scmutil,
    util,
)
from mercurial.utils import (
    procutil,
    stringutil,
)

cmdtable = {}
command = registrar.command(cmdtable)

configtable = {}
configitem = registrar.configitem(configtable)

configitem('extdiff', br'opts\..*',
    default='',
    generic=True,
)

configitem('diff-tools', br'.*\.diffargs$',
    default=None,
    generic=True,
)

# Note for extension authors: ONLY specify testedwith = 'ships-with-hg-core' for
# extensions which SHIP WITH MERCURIAL. Non-mainline extensions should
# be specifying the version(s) of Mercurial they are tested with, or
# leave the attribute unspecified.
testedwith = 'ships-with-hg-core'

def snapshot(ui, repo, files, node, tmproot, listsubrepos):
    '''snapshot files as of some revision
    if not using snapshot, -I/-X does not work and recursive diff
    in tools like kdiff3 and meld displays too many files.'''
    dirname = os.path.basename(repo.root)
    if dirname == "":
        dirname = "root"
    if node is not None:
        dirname = '%s.%s' % (dirname, short(node))
    base = os.path.join(tmproot, dirname)
    os.mkdir(base)
    fnsandstat = []

    if node is not None:
        ui.note(_('making snapshot of %d files from rev %s\n') %
                (len(files), short(node)))
    else:
        ui.note(_('making snapshot of %d files from working directory\n') %
            (len(files)))

    if files:
        repo.ui.setconfig("ui", "archivemeta", False)

        archival.archive(repo, base, node, 'files',
                         matchfn=scmutil.matchfiles(repo, files),
                         subrepos=listsubrepos)

        for fn in sorted(files):
            wfn = util.pconvert(fn)
            ui.note('  %s\n' % wfn)

            if node is None:
                dest = os.path.join(base, wfn)

                fnsandstat.append((dest, repo.wjoin(fn), os.lstat(dest)))
    return dirname, fnsandstat

def dodiff(ui, repo, cmdline, pats, opts):
    '''Do the actual diff:

    - copy to a temp structure if diffing 2 internal revisions
    - copy to a temp structure if diffing working revision with
      another one and more than 1 file is changed
    - just invoke the diff for a single file in the working dir
    '''

    revs = opts.get('rev')
    change = opts.get('change')
    do3way = '$parent2' in cmdline

    if revs and change:
        msg = _('cannot specify --rev and --change at the same time')
        raise error.Abort(msg)
    elif change:
        ctx2 = scmutil.revsingle(repo, change, None)
        ctx1a, ctx1b = ctx2.p1(), ctx2.p2()
    else:
        ctx1a, ctx2 = scmutil.revpair(repo, revs)
        if not revs:
            ctx1b = repo[None].p2()
        else:
            ctx1b = repo[nullid]

    node1a = ctx1a.node()
    node1b = ctx1b.node()
    node2 = ctx2.node()

    # Disable 3-way merge if there is only one parent
    if do3way:
        if node1b == nullid:
            do3way = False

    subrepos=opts.get('subrepos')

    matcher = scmutil.match(repo[node2], pats, opts)

    if opts.get('patch'):
        if subrepos:
            raise error.Abort(_('--patch cannot be used with --subrepos'))
        if node2 is None:
            raise error.Abort(_('--patch requires two revisions'))
    else:
        mod_a, add_a, rem_a = map(set, repo.status(node1a, node2, matcher,
                                                   listsubrepos=subrepos)[:3])
        if do3way:
            mod_b, add_b, rem_b = map(set,
                                      repo.status(node1b, node2, matcher,
                                                  listsubrepos=subrepos)[:3])
        else:
            mod_b, add_b, rem_b = set(), set(), set()
        modadd = mod_a | add_a | mod_b | add_b
        common = modadd | rem_a | rem_b
        if not common:
            return 0

    tmproot = pycompat.mkdtemp(prefix='extdiff.')
    try:
        if not opts.get('patch'):
            # Always make a copy of node1a (and node1b, if applicable)
            dir1a_files = mod_a | rem_a | ((mod_b | add_b) - add_a)
            dir1a = snapshot(ui, repo, dir1a_files, node1a, tmproot,
                             subrepos)[0]
            rev1a = '@%d' % repo[node1a].rev()
            if do3way:
                dir1b_files = mod_b | rem_b | ((mod_a | add_a) - add_b)
                dir1b = snapshot(ui, repo, dir1b_files, node1b, tmproot,
                                 subrepos)[0]
                rev1b = '@%d' % repo[node1b].rev()
            else:
                dir1b = None
                rev1b = ''

            fnsandstat = []

            # If node2 in not the wc or there is >1 change, copy it
            dir2root = ''
            rev2 = ''
            if node2:
                dir2 = snapshot(ui, repo, modadd, node2, tmproot, subrepos)[0]
                rev2 = '@%d' % repo[node2].rev()
            elif len(common) > 1:
                #we only actually need to get the files to copy back to
                #the working dir in this case (because the other cases
                #are: diffing 2 revisions or single file -- in which case
                #the file is already directly passed to the diff tool).
                dir2, fnsandstat = snapshot(ui, repo, modadd, None, tmproot,
                                            subrepos)
            else:
                # This lets the diff tool open the changed file directly
                dir2 = ''
                dir2root = repo.root

            label1a = rev1a
            label1b = rev1b
            label2 = rev2

            # If only one change, diff the files instead of the directories
            # Handle bogus modifies correctly by checking if the files exist
            if len(common) == 1:
                common_file = util.localpath(common.pop())
                dir1a = os.path.join(tmproot, dir1a, common_file)
                label1a = common_file + rev1a
                if not os.path.isfile(dir1a):
                    dir1a = os.devnull
                if do3way:
                    dir1b = os.path.join(tmproot, dir1b, common_file)
                    label1b = common_file + rev1b
                    if not os.path.isfile(dir1b):
                        dir1b = os.devnull
                dir2 = os.path.join(dir2root, dir2, common_file)
                label2 = common_file + rev2
        else:
            template = 'hg-%h.patch'
            with formatter.nullformatter(ui, 'extdiff', {}) as fm:
                cmdutil.export(repo, [repo[node1a].rev(), repo[node2].rev()],
                               fm,
                               fntemplate=repo.vfs.reljoin(tmproot, template),
                               match=matcher)
            label1a = cmdutil.makefilename(repo[node1a], template)
            label2 = cmdutil.makefilename(repo[node2], template)
            dir1a = repo.vfs.reljoin(tmproot, label1a)
            dir2 = repo.vfs.reljoin(tmproot, label2)
            dir1b = None
            label1b = None
            fnsandstat = []

        # Function to quote file/dir names in the argument string.
        # When not operating in 3-way mode, an empty string is
        # returned for parent2
        replace = {'parent': dir1a, 'parent1': dir1a, 'parent2': dir1b,
                   'plabel1': label1a, 'plabel2': label1b,
                   'clabel': label2, 'child': dir2,
                   'root': repo.root}
        def quote(match):
            pre = match.group(2)
            key = match.group(3)
            if not do3way and key == 'parent2':
                return pre
            return pre + procutil.shellquote(replace[key])

        # Match parent2 first, so 'parent1?' will match both parent1 and parent
        regex = (br'''(['"]?)([^\s'"$]*)'''
                 br'\$(parent2|parent1?|child|plabel1|plabel2|clabel|root)\1')
        if not do3way and not re.search(regex, cmdline):
            cmdline += ' $parent1 $child'
        cmdline = re.sub(regex, quote, cmdline)

        ui.debug('running %r in %s\n' % (pycompat.bytestr(cmdline), tmproot))
        ui.system(cmdline, cwd=tmproot, blockedtag='extdiff')

        for copy_fn, working_fn, st in fnsandstat:
            cpstat = os.lstat(copy_fn)
            # Some tools copy the file and attributes, so mtime may not detect
            # all changes.  A size check will detect more cases, but not all.
            # The only certain way to detect every case is to diff all files,
            # which could be expensive.
            # copyfile() carries over the permission, so the mode check could
            # be in an 'elif' branch, but for the case where the file has
            # changed without affecting mtime or size.
            if (cpstat[stat.ST_MTIME] != st[stat.ST_MTIME]
                or cpstat.st_size != st.st_size
                or (cpstat.st_mode & 0o100) != (st.st_mode & 0o100)):
                ui.debug('file changed while diffing. '
                         'Overwriting: %s (src: %s)\n' % (working_fn, copy_fn))
                util.copyfile(copy_fn, working_fn)

        return 1
    finally:
        ui.note(_('cleaning up temp directory\n'))
        shutil.rmtree(tmproot)

extdiffopts = [
    ('o', 'option', [],
     _('pass option to comparison program'), _('OPT')),
    ('r', 'rev', [], _('revision'), _('REV')),
    ('c', 'change', '', _('change made by revision'), _('REV')),
    ('', 'patch', None, _('compare patches for two revisions'))
    ] + cmdutil.walkopts + cmdutil.subrepoopts

@command('extdiff',
    [('p', 'program', '', _('comparison program to run'), _('CMD')),
     ] + extdiffopts,
    _('hg extdiff [OPT]... [FILE]...'),
    inferrepo=True)
def extdiff(ui, repo, *pats, **opts):
    '''use external program to diff repository (or selected files)

    Show differences between revisions for the specified files, using
    an external program. The default program used is diff, with
    default options "-Npru".

    To select a different program, use the -p/--program option. The
    program will be passed the names of two directories to compare. To
    pass additional options to the program, use -o/--option. These
    will be passed before the names of the directories to compare.

    When two revision arguments are given, then changes are shown
    between those revisions. If only one revision is specified then
    that revision is compared to the working directory, and, when no
    revisions are specified, the working directory files are compared
    to its parent.'''
    opts = pycompat.byteskwargs(opts)
    program = opts.get('program')
    option = opts.get('option')
    if not program:
        program = 'diff'
        option = option or ['-Npru']
    cmdline = ' '.join(map(procutil.shellquote, [program] + option))
    return dodiff(ui, repo, cmdline, pats, opts)

class savedcmd(object):
    """use external program to diff repository (or selected files)

    Show differences between revisions for the specified files, using
    the following program::

        %(path)s

    When two revision arguments are given, then changes are shown
    between those revisions. If only one revision is specified then
    that revision is compared to the working directory, and, when no
    revisions are specified, the working directory files are compared
    to its parent.
    """

    def __init__(self, path, cmdline):
        # We can't pass non-ASCII through docstrings (and path is
        # in an unknown encoding anyway)
        docpath = stringutil.escapestr(path)
        self.__doc__ %= {r'path': pycompat.sysstr(stringutil.uirepr(docpath))}
        self._cmdline = cmdline

    def __call__(self, ui, repo, *pats, **opts):
        opts = pycompat.byteskwargs(opts)
        options = ' '.join(map(procutil.shellquote, opts['option']))
        if options:
            options = ' ' + options
        return dodiff(ui, repo, self._cmdline + options, pats, opts)

def uisetup(ui):
    for cmd, path in ui.configitems('extdiff'):
        path = util.expandpath(path)
        if cmd.startswith('cmd.'):
            cmd = cmd[4:]
            if not path:
                path = procutil.findexe(cmd)
                if path is None:
                    path = filemerge.findexternaltool(ui, cmd) or cmd
            diffopts = ui.config('extdiff', 'opts.' + cmd)
            cmdline = procutil.shellquote(path)
            if diffopts:
                cmdline += ' ' + diffopts
        elif cmd.startswith('opts.'):
            continue
        else:
            if path:
                # case "cmd = path opts"
                cmdline = path
                diffopts = len(pycompat.shlexsplit(cmdline)) > 1
            else:
                # case "cmd ="
                path = procutil.findexe(cmd)
                if path is None:
                    path = filemerge.findexternaltool(ui, cmd) or cmd
                cmdline = procutil.shellquote(path)
                diffopts = False
        # look for diff arguments in [diff-tools] then [merge-tools]
        if not diffopts:
            args = ui.config('diff-tools', cmd+'.diffargs') or \
                   ui.config('merge-tools', cmd+'.diffargs')
            if args:
                cmdline += ' ' + args
        command(cmd, extdiffopts[:], _('hg %s [OPTION]... [FILE]...') % cmd,
                inferrepo=True)(savedcmd(path, cmdline))

# tell hggettext to extract docstrings from these functions:
i18nfunctions = [savedcmd]