Mercurial > hg
view tests/test-linelog.py @ 42621:99ebde4fec99
commit: improve the files field of changelog for merges
Currently, the files list of merge commits repeats all the deletions
(either actual deletions, or files that got renamed) that happened
between base and p2 of the merge. If p2 is the main branch, the list
can easily be much bigger than the change being merged.
This results in various problems worth improving:
- changelog is bigger than necessary
- `hg log directory` lists many unrelated merge commits, and `hg log
-v -r commit` frequently fills multiple screens worth of files
- it possibly slows down adjustlinkrev, by forcing it to read more
manifests, and that function can certainly be a bottleneck
- the server side of pulls can waste a lot of time simply opening the
filelogs for pointless files (the constant factors for opening even
a tiny filelog is apparently pretty bad)
So stop listing such files as described in the code. Impacted merge
commits and their descendants get a different hash than they would
have without this. This doesn't seem problematic, except for
convert. The previous commit helped with that in the hg->hg case (but
if you do svn->hg twice from scratch, hashes can still change).
The rest of the description is numbers. I don't have much to report,
because recreating the files list of existing repositories is not
easy:
- debugupgradeformat and bundle/unbundle don't recreate the list
- export/import tends to choke quickly applying patches or on
description that contain diffs,
- merge commits from the convert extension don't have the right files
list for reasons orthogonal to the current commit
- replaying the merge with hg update/hg merge/hg revert --all/hg
commit can end up failing in hg revert
- I wasn't sure that using debugsetparents + debugrebuilddirstate
would really build the right thing
I measured commit time before and after this change, in a case with no
files filtered out, several files filtered out (no difference) and 5k
files filtered out (+1% time).
Recreating the 100 more recent merges in a private repo, the
concatenated uncompressed files lists goes from 1.12MB to
0.52MB. Excluding 3 merges that are not representative, then the size
goes from 570k to 15k.
I converted part of mozilla-central, and observed file list shrinking
quite a bit too, starting at the very first merge, 733641d9feaf, going
from 550 files to 10 files (although they have relatively few merges,
so they probably wouldn't care).
Differential Revision: https://phab.mercurial-scm.org/D6613
author | Valentin Gatien-Baron <valentin.gatienbaron@gmail.com> |
---|---|
date | Tue, 02 Jul 2019 12:59:58 -0400 |
parents | 876494fd967d |
children | 2372284d9457 |
line wrap: on
line source
from __future__ import absolute_import, print_function import difflib import random import unittest from mercurial import linelog vecratio = 3 # number of replacelines / number of replacelines_vec maxlinenum = 0xffffff maxb1 = 0xffffff maxdeltaa = 10 maxdeltab = 10 def _genedits(seed, endrev): lines = [] random.seed(seed) for rev in range(0, endrev): n = len(lines) a1 = random.randint(0, n) a2 = random.randint(a1, min(n, a1 + maxdeltaa)) b1 = random.randint(0, maxb1) b2 = random.randint(b1, b1 + maxdeltab) usevec = not bool(random.randint(0, vecratio)) if usevec: blines = [(random.randint(0, rev), random.randint(0, maxlinenum)) for _ in range(b1, b2)] else: blines = [(rev, bidx) for bidx in range(b1, b2)] lines[a1:a2] = blines yield lines, rev, a1, a2, b1, b2, blines, usevec class linelogtests(unittest.TestCase): def testlinelogencodedecode(self): program = [linelog._eof(0, 0), linelog._jge(41, 42), linelog._jump(0, 43), linelog._eof(0, 0), linelog._jl(44, 45), linelog._line(46, 47), ] ll = linelog.linelog(program, maxrev=100) enc = ll.encode() # round-trips okay self.assertEqual(linelog.linelog.fromdata(enc)._program, ll._program) self.assertEqual(linelog.linelog.fromdata(enc), ll) # This encoding matches the encoding used by hg-experimental's # linelog file, or is supposed to if it doesn't. self.assertEqual(enc, (b'\x00\x00\x01\x90\x00\x00\x00\x06' b'\x00\x00\x00\xa4\x00\x00\x00*' b'\x00\x00\x00\x00\x00\x00\x00+' b'\x00\x00\x00\x00\x00\x00\x00\x00' b'\x00\x00\x00\xb1\x00\x00\x00-' b'\x00\x00\x00\xba\x00\x00\x00/')) def testsimpleedits(self): ll = linelog.linelog() # Initial revision: add lines 0, 1, and 2 ll.replacelines(1, 0, 0, 0, 3) self.assertEqual([(l.rev, l.linenum) for l in ll.annotate(1)], [(1, 0), (1, 1), (1, 2), ]) # Replace line 1 with a new line ll.replacelines(2, 1, 2, 1, 2) self.assertEqual([(l.rev, l.linenum) for l in ll.annotate(2)], [(1, 0), (2, 1), (1, 2), ]) # delete a line out of 2 ll.replacelines(3, 1, 2, 0, 0) self.assertEqual([(l.rev, l.linenum) for l in ll.annotate(3)], [(1, 0), (1, 2), ]) # annotation of 1 is unchanged self.assertEqual([(l.rev, l.linenum) for l in ll.annotate(1)], [(1, 0), (1, 1), (1, 2), ]) ll.annotate(3) # set internal state to revision 3 start = ll.getoffset(0) end = ll.getoffset(1) self.assertEqual(ll.getalllines(start, end), [ (1, 0), (2, 1), (1, 1), ]) self.assertEqual(ll.getalllines(), [ (1, 0), (2, 1), (1, 1), (1, 2), ]) def testparseclinelogfile(self): # This data is what the replacements in testsimpleedits # produce when fed to the original linelog.c implementation. data = (b'\x00\x00\x00\x0c\x00\x00\x00\x0f' b'\x00\x00\x00\x00\x00\x00\x00\x02' b'\x00\x00\x00\x05\x00\x00\x00\x06' b'\x00\x00\x00\x06\x00\x00\x00\x00' b'\x00\x00\x00\x00\x00\x00\x00\x07' b'\x00\x00\x00\x06\x00\x00\x00\x02' b'\x00\x00\x00\x00\x00\x00\x00\x00' b'\x00\x00\x00\t\x00\x00\x00\t' b'\x00\x00\x00\x00\x00\x00\x00\x0c' b'\x00\x00\x00\x08\x00\x00\x00\x05' b'\x00\x00\x00\x06\x00\x00\x00\x01' b'\x00\x00\x00\x00\x00\x00\x00\x05' b'\x00\x00\x00\x0c\x00\x00\x00\x05' b'\x00\x00\x00\n\x00\x00\x00\x01' b'\x00\x00\x00\x00\x00\x00\x00\t') llc = linelog.linelog.fromdata(data) self.assertEqual([(l.rev, l.linenum) for l in llc.annotate(1)], [(1, 0), (1, 1), (1, 2), ]) self.assertEqual([(l.rev, l.linenum) for l in llc.annotate(2)], [(1, 0), (2, 1), (1, 2), ]) self.assertEqual([(l.rev, l.linenum) for l in llc.annotate(3)], [(1, 0), (1, 2), ]) # Check we emit the same bytecode. ll = linelog.linelog() # Initial revision: add lines 0, 1, and 2 ll.replacelines(1, 0, 0, 0, 3) # Replace line 1 with a new line ll.replacelines(2, 1, 2, 1, 2) # delete a line out of 2 ll.replacelines(3, 1, 2, 0, 0) diff = '\n ' + '\n '.join(difflib.unified_diff( ll.debugstr().splitlines(), llc.debugstr().splitlines(), 'python', 'c', lineterm='')) self.assertEqual(ll._program, llc._program, 'Program mismatch: ' + diff) # Done as a secondary step so we get a better result if the # program is where the mismatch is. self.assertEqual(ll, llc) self.assertEqual(ll.encode(), data) def testanothersimplecase(self): ll = linelog.linelog() ll.replacelines(3, 0, 0, 0, 2) ll.replacelines(4, 0, 2, 0, 0) self.assertEqual([(l.rev, l.linenum) for l in ll.annotate(4)], []) self.assertEqual([(l.rev, l.linenum) for l in ll.annotate(3)], [(3, 0), (3, 1)]) # rev 2 is empty because contents were only ever introduced in rev 3 self.assertEqual([(l.rev, l.linenum) for l in ll.annotate(2)], []) def testrandomedits(self): # Inspired by original linelog tests. seed = random.random() numrevs = 2000 ll = linelog.linelog() # Populate linelog for lines, rev, a1, a2, b1, b2, blines, usevec in _genedits( seed, numrevs): if usevec: ll.replacelines_vec(rev, a1, a2, blines) else: ll.replacelines(rev, a1, a2, b1, b2) ar = ll.annotate(rev) self.assertEqual(ll.annotateresult, lines) # Verify we can get back these states by annotating each rev for lines, rev, a1, a2, b1, b2, blines, usevec in _genedits( seed, numrevs): ar = ll.annotate(rev) self.assertEqual([(l.rev, l.linenum) for l in ar], lines) def testinfinitebadprogram(self): ll = linelog.linelog.fromdata( b'\x00\x00\x00\x00\x00\x00\x00\x02' # header b'\x00\x00\x00\x00\x00\x00\x00\x01' # JUMP to self ) with self.assertRaises(linelog.LineLogError): # should not be an infinite loop and raise ll.annotate(1) if __name__ == '__main__': import silenttestrunner silenttestrunner.main(__name__)