Mercurial > hg
view mercurial/simplemerge.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 | e7aa113b14f7 |
children | aaad36b88298 |
line wrap: on
line source
# Copyright (C) 2004, 2005 Canonical Ltd # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2 of the License, or # (at your option) any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, see <http://www.gnu.org/licenses/>. # mbp: "you know that thing where cvs gives you conflict markers?" # s: "i hate that." from __future__ import absolute_import from .i18n import _ from . import ( error, mdiff, pycompat, ) from .utils import ( stringutil, ) class CantReprocessAndShowBase(Exception): pass def intersect(ra, rb): """Given two ranges return the range where they intersect or None. >>> intersect((0, 10), (0, 6)) (0, 6) >>> intersect((0, 10), (5, 15)) (5, 10) >>> intersect((0, 10), (10, 15)) >>> intersect((0, 9), (10, 15)) >>> intersect((0, 9), (7, 15)) (7, 9) """ assert ra[0] <= ra[1] assert rb[0] <= rb[1] sa = max(ra[0], rb[0]) sb = min(ra[1], rb[1]) if sa < sb: return sa, sb else: return None def compare_range(a, astart, aend, b, bstart, bend): """Compare a[astart:aend] == b[bstart:bend], without slicing. """ if (aend - astart) != (bend - bstart): return False for ia, ib in zip(pycompat.xrange(astart, aend), pycompat.xrange(bstart, bend)): if a[ia] != b[ib]: return False else: return True class Merge3Text(object): """3-way merge of texts. Given strings BASE, OTHER, THIS, tries to produce a combined text incorporating the changes from both BASE->OTHER and BASE->THIS.""" def __init__(self, basetext, atext, btext, base=None, a=None, b=None): self.basetext = basetext self.atext = atext self.btext = btext if base is None: base = mdiff.splitnewlines(basetext) if a is None: a = mdiff.splitnewlines(atext) if b is None: b = mdiff.splitnewlines(btext) self.base = base self.a = a self.b = b def merge_lines(self, name_a=None, name_b=None, name_base=None, start_marker='<<<<<<<', mid_marker='=======', end_marker='>>>>>>>', base_marker=None, localorother=None, minimize=False): """Return merge in cvs-like form. """ self.conflicts = False newline = '\n' if len(self.a) > 0: if self.a[0].endswith('\r\n'): newline = '\r\n' elif self.a[0].endswith('\r'): newline = '\r' if name_a and start_marker: start_marker = start_marker + ' ' + name_a if name_b and end_marker: end_marker = end_marker + ' ' + name_b if name_base and base_marker: base_marker = base_marker + ' ' + name_base merge_regions = self.merge_regions() if minimize: merge_regions = self.minimize(merge_regions) for t in merge_regions: what = t[0] if what == 'unchanged': for i in range(t[1], t[2]): yield self.base[i] elif what == 'a' or what == 'same': for i in range(t[1], t[2]): yield self.a[i] elif what == 'b': for i in range(t[1], t[2]): yield self.b[i] elif what == 'conflict': if localorother == 'local': for i in range(t[3], t[4]): yield self.a[i] elif localorother == 'other': for i in range(t[5], t[6]): yield self.b[i] else: self.conflicts = True if start_marker is not None: yield start_marker + newline for i in range(t[3], t[4]): yield self.a[i] if base_marker is not None: yield base_marker + newline for i in range(t[1], t[2]): yield self.base[i] if mid_marker is not None: yield mid_marker + newline for i in range(t[5], t[6]): yield self.b[i] if end_marker is not None: yield end_marker + newline else: raise ValueError(what) def merge_groups(self): """Yield sequence of line groups. Each one is a tuple: 'unchanged', lines Lines unchanged from base 'a', lines Lines taken from a 'same', lines Lines taken from a (and equal to b) 'b', lines Lines taken from b 'conflict', base_lines, a_lines, b_lines Lines from base were changed to either a or b and conflict. """ for t in self.merge_regions(): what = t[0] if what == 'unchanged': yield what, self.base[t[1]:t[2]] elif what == 'a' or what == 'same': yield what, self.a[t[1]:t[2]] elif what == 'b': yield what, self.b[t[1]:t[2]] elif what == 'conflict': yield (what, self.base[t[1]:t[2]], self.a[t[3]:t[4]], self.b[t[5]:t[6]]) else: raise ValueError(what) def merge_regions(self): """Return sequences of matching and conflicting regions. This returns tuples, where the first value says what kind we have: 'unchanged', start, end Take a region of base[start:end] 'same', astart, aend b and a are different from base but give the same result 'a', start, end Non-clashing insertion from a[start:end] 'conflict', zstart, zend, astart, aend, bstart, bend Conflict between a and b, with z as common ancestor Method is as follows: The two sequences align only on regions which match the base and both descendants. These are found by doing a two-way diff of each one against the base, and then finding the intersections between those regions. These "sync regions" are by definition unchanged in both and easily dealt with. The regions in between can be in any of three cases: conflicted, or changed on only one side. """ # section a[0:ia] has been disposed of, etc iz = ia = ib = 0 for region in self.find_sync_regions(): zmatch, zend, amatch, aend, bmatch, bend = region #print 'match base [%d:%d]' % (zmatch, zend) matchlen = zend - zmatch assert matchlen >= 0 assert matchlen == (aend - amatch) assert matchlen == (bend - bmatch) len_a = amatch - ia len_b = bmatch - ib len_base = zmatch - iz assert len_a >= 0 assert len_b >= 0 assert len_base >= 0 #print 'unmatched a=%d, b=%d' % (len_a, len_b) if len_a or len_b: # try to avoid actually slicing the lists equal_a = compare_range(self.a, ia, amatch, self.base, iz, zmatch) equal_b = compare_range(self.b, ib, bmatch, self.base, iz, zmatch) same = compare_range(self.a, ia, amatch, self.b, ib, bmatch) if same: yield 'same', ia, amatch elif equal_a and not equal_b: yield 'b', ib, bmatch elif equal_b and not equal_a: yield 'a', ia, amatch elif not equal_a and not equal_b: yield 'conflict', iz, zmatch, ia, amatch, ib, bmatch else: raise AssertionError("can't handle a=b=base but unmatched") ia = amatch ib = bmatch iz = zmatch # if the same part of the base was deleted on both sides # that's OK, we can just skip it. if matchlen > 0: assert ia == amatch assert ib == bmatch assert iz == zmatch yield 'unchanged', zmatch, zend iz = zend ia = aend ib = bend def minimize(self, merge_regions): """Trim conflict regions of lines where A and B sides match. Lines where both A and B have made the same changes at the beginning or the end of each merge region are eliminated from the conflict region and are instead considered the same. """ for region in merge_regions: if region[0] != "conflict": yield region continue issue, z1, z2, a1, a2, b1, b2 = region alen = a2 - a1 blen = b2 - b1 # find matches at the front ii = 0 while ii < alen and ii < blen and \ self.a[a1 + ii] == self.b[b1 + ii]: ii += 1 startmatches = ii # find matches at the end ii = 0 while ii < alen and ii < blen and \ self.a[a2 - ii - 1] == self.b[b2 - ii - 1]: ii += 1 endmatches = ii if startmatches > 0: yield 'same', a1, a1 + startmatches yield ('conflict', z1, z2, a1 + startmatches, a2 - endmatches, b1 + startmatches, b2 - endmatches) if endmatches > 0: yield 'same', a2 - endmatches, a2 def find_sync_regions(self): """Return a list of sync regions, where both descendants match the base. Generates a list of (base1, base2, a1, a2, b1, b2). There is always a zero-length sync region at the end of all the files. """ ia = ib = 0 amatches = mdiff.get_matching_blocks(self.basetext, self.atext) bmatches = mdiff.get_matching_blocks(self.basetext, self.btext) len_a = len(amatches) len_b = len(bmatches) sl = [] while ia < len_a and ib < len_b: abase, amatch, alen = amatches[ia] bbase, bmatch, blen = bmatches[ib] # there is an unconflicted block at i; how long does it # extend? until whichever one ends earlier. i = intersect((abase, abase + alen), (bbase, bbase + blen)) if i: intbase = i[0] intend = i[1] intlen = intend - intbase # found a match of base[i[0], i[1]]; this may be less than # the region that matches in either one assert intlen <= alen assert intlen <= blen assert abase <= intbase assert bbase <= intbase asub = amatch + (intbase - abase) bsub = bmatch + (intbase - bbase) aend = asub + intlen bend = bsub + intlen assert self.base[intbase:intend] == self.a[asub:aend], \ (self.base[intbase:intend], self.a[asub:aend]) assert self.base[intbase:intend] == self.b[bsub:bend] sl.append((intbase, intend, asub, aend, bsub, bend)) # advance whichever one ends first in the base text if (abase + alen) < (bbase + blen): ia += 1 else: ib += 1 intbase = len(self.base) abase = len(self.a) bbase = len(self.b) sl.append((intbase, intbase, abase, abase, bbase, bbase)) return sl def find_unconflicted(self): """Return a list of ranges in base that are not conflicted.""" am = mdiff.get_matching_blocks(self.basetext, self.atext) bm = mdiff.get_matching_blocks(self.basetext, self.btext) unc = [] while am and bm: # there is an unconflicted block at i; how long does it # extend? until whichever one ends earlier. a1 = am[0][0] a2 = a1 + am[0][2] b1 = bm[0][0] b2 = b1 + bm[0][2] i = intersect((a1, a2), (b1, b2)) if i: unc.append(i) if a2 < b2: del am[0] else: del bm[0] return unc def _verifytext(text, path, ui, opts): """verifies that text is non-binary (unless opts[text] is passed, then we just warn)""" if stringutil.binary(text): msg = _("%s looks like a binary file.") % path if not opts.get('quiet'): ui.warn(_('warning: %s\n') % msg) if not opts.get('text'): raise error.Abort(msg) return text def _picklabels(defaults, overrides): if len(overrides) > 3: raise error.Abort(_("can only specify three labels.")) result = defaults[:] for i, override in enumerate(overrides): result[i] = override return result def simplemerge(ui, localctx, basectx, otherctx, **opts): """Performs the simplemerge algorithm. The merged result is written into `localctx`. """ opts = pycompat.byteskwargs(opts) def readctx(ctx): # Merges were always run in the working copy before, which means # they used decoded data, if the user defined any repository # filters. # # Maintain that behavior today for BC, though perhaps in the future # it'd be worth considering whether merging encoded data (what the # repository usually sees) might be more useful. return _verifytext(ctx.decodeddata(), ctx.path(), ui, opts) mode = opts.get('mode','merge') name_a, name_b, name_base = None, None, None if mode != 'union': name_a, name_b, name_base = _picklabels([localctx.path(), otherctx.path(), None], opts.get('label', [])) try: localtext = readctx(localctx) basetext = readctx(basectx) othertext = readctx(otherctx) except error.Abort: return 1 m3 = Merge3Text(basetext, localtext, othertext) extrakwargs = { "localorother": opts.get("localorother", None), 'minimize': True, } if mode == 'union': extrakwargs['start_marker'] = None extrakwargs['mid_marker'] = None extrakwargs['end_marker'] = None elif name_base is not None: extrakwargs['base_marker'] = '|||||||' extrakwargs['name_base'] = name_base extrakwargs['minimize'] = False mergedtext = "" for line in m3.merge_lines(name_a=name_a, name_b=name_b, **pycompat.strkwargs(extrakwargs)): if opts.get('print'): ui.fout.write(line) else: mergedtext += line if not opts.get('print'): localctx.write(mergedtext, localctx.flags()) if m3.conflicts and not mode == 'union': return 1