sparse-revlog: implement algorithm to write sparse delta chains (
issue5480)
The classic behavior of revlog._isgooddeltainfo is to consider the span size
of the whole delta chain, and limit it to 4 * textlen.
Once sparse-revlog writing is allowed (and enforced with a requirement),
revlog._isgooddeltainfo considers the span of the largest chunk as the
distance used in the verification, instead of using the span of the whole
delta chain.
In order to compute the span of the largest chunk, we need to slice into
chunks a chain with the new revision at the top of the revlog, and take the
maximal span of these chunks. The sparse read density is a parameter to the
slicing, as it will stop when the global read density reaches this threshold.
For instance, a density of 50% means that 2 of 4 read bytes are actually used
for the reconstruction of the revision (the others are part of other chains).
This allows a new revision to be potentially stored with a diff against
another revision anywhere in the history, instead of forcing it in the last 4
* textlen. The result is a much better compression on repositories that have
many concurrent branches. Here are a comparison between using deltas from
current upstream (aggressive-merge-deltas on by default) and deltas from a
sparse-revlog
Comparison of `.hg/store/` size:
mercurial (6.74% merges):
before: 46,831,873 bytes
after: 46,795,992 bytes (no relevant change)
pypy (8.30% merges):
before: 333,524,651 bytes
after: 308,417,511 bytes -8%
netbeans (34.21% merges):
before: 1,141,847,554 bytes
after: 1,131,093,161 bytes -1%
mozilla-central (4.84% merges):
before: 2,344,248,850 bytes
after: 2,328,459,258 bytes -1%
large-private-repo-A (merge 19.73%)
before: 41,510,550,163 bytes
after: 8,121,763,428 bytes -80%
large-private-repo-B (23.77%)
before: 58,702,221,709 bytes
after: 8,351,588,828 bytes -76%
Comparison of `00manifest.d` size:
mercurial (6.74% merges):
before: 6,143,044 bytes
after: 6,107,163 bytes
pypy (8.30% merges):
before: 52,941,780 bytes
after: 27,834,082 bytes -48%
netbeans (34.21% merges):
before: 130,088,982 bytes
after: 119,337,636 bytes -10%
mozilla-central (4.84% merges):
before: 215,096,339 bytes
after: 199,496,863 bytes -8%
large-private-repo-A (merge 19.73%)
before: 33,725,285,081 bytes
after: 390,302,545 bytes -99%
large-private-repo-B (23.77%)
before: 49,457,701,645 bytes
after: 1,366,752,187 bytes -97%
The better delta chains provide a performance boost in relevant repositories:
pypy, bundling 1000 revisions:
before: 1.670s
after: 1.149s -31%
Unbundling got a bit slower. probably because the sparse algorithm is still
pure
python.
pypy, unbundling 1000 revisions:
before: 4.062s
after: 4.507s +10%
Performance of bundle/unbundle in repository with few concurrent branches (eg:
mercurial) are unaffected.
No significant differences have been noticed then timing `hg push` and `hg
pull` locally. More state timings are being gathered.
Same as for aggressive-merge-delta, better delta comes with longer delta
chains. Longer chains have a performance impact. For example. The length of
the chain needed to get the manifest of pypy's tip moves from 82 item to 1929
items. This moves the restore time from 3.88ms to 11.3ms.
Delta chain length is an independent issue that affects repository without
this changes. It will be dealt with independently.
No significant differences have been observed on repositories where
`sparse-revlog` have not much effect (mercurial, unity, netbeans). On pypy,
small differences have been observed on some operation affected by delta chain
building and retrieval.
pypy, perfmanifest
before: 0.006162s
after: 0.017899s +190%
pypy, commit:
before: 0.382
after: 0.376 -1%
pypy, status:
before: 0.157
after: 0.168 +7%
More comprehensive and stable timing comparisons are in progress.
$ cat > loop.py <<EOF
> from __future__ import absolute_import
> import time
> from mercurial import commands, registrar
>
> cmdtable = {}
> command = registrar.command(cmdtable)
>
> class incrementingtime(object):
> def __init__(self):
> self._time = 0.0
> def __call__(self):
> self._time += 0.25
> return self._time
> time.time = incrementingtime()
>
> @command(b'loop',
> [(b'', b'total', b'', b'override for total'),
> (b'', b'nested', False, b'show nested results'),
> (b'', b'parallel', False, b'show parallel sets of results')],
> b'hg loop LOOPS',
> norepo=True)
> def loop(ui, loops, **opts):
> loops = int(loops)
> total = None
> if loops >= 0:
> total = loops
> if opts.get('total', None):
> total = int(opts.get('total'))
> nested = False
> if opts.get('nested', None):
> nested = True
> loops = abs(loops)
>
> progress = ui.makeprogress(topiclabel, unit='loopnum', total=total)
> other = ui.makeprogress('other', unit='othernum', total=total)
> for i in range(loops):
> progress.update(i, item=getloopitem(i))
> if opts.get('parallel'):
> other.update(i, item='other.%d' % i)
> if nested:
> nested_steps = 2
> if i and i % 4 == 0:
> nested_steps = 5
> nested = ui.makeprogress('nested', unit='nestnum',
> total=nested_steps)
> for j in range(nested_steps):
> nested.update(j, item='nested.%d' % j)
> nested.complete()
> progress.complete()
>
> topiclabel = 'loop'
> def getloopitem(i):
> return 'loop.%d' % i
>
> EOF
$ cp $HGRCPATH $HGRCPATH.orig
$ echo "[extensions]" >> $HGRCPATH
$ echo "progress=" >> $HGRCPATH
$ echo "loop=`pwd`/loop.py" >> $HGRCPATH
$ echo "[progress]" >> $HGRCPATH
$ echo "format = topic bar number" >> $HGRCPATH
$ echo "assume-tty=1" >> $HGRCPATH
$ echo "width=60" >> $HGRCPATH
test default params, display nothing because of delay
$ hg -y loop 3
$ echo "delay=0" >> $HGRCPATH
$ echo "refresh=0" >> $HGRCPATH
test with delay=0, refresh=0
$ hg -y loop 3
\r (no-eol) (esc)
loop [ ] 0/3\r (no-eol) (esc)
loop [===============> ] 1/3\r (no-eol) (esc)
loop [===============================> ] 2/3\r (no-eol) (esc)
\r (no-eol) (esc)
no progress with --quiet
$ hg -y loop 3 --quiet
test plain mode exception
$ HGPLAINEXCEPT=progress hg -y loop 1
\r (no-eol) (esc)
loop [ ] 0/1\r (no-eol) (esc)
\r (no-eol) (esc)
test nested short-lived topics (which shouldn't display with nestdelay):
$ hg -y loop 3 --nested
\r (no-eol) (esc)
loop [ ] 0/3\r (no-eol) (esc)
loop [===============> ] 1/3\r (no-eol) (esc)
loop [===============================> ] 2/3\r (no-eol) (esc)
\r (no-eol) (esc)
Test nested long-lived topic which has the same name as a short-lived
peer. We shouldn't get stuck showing the short-lived inner steps, and
should go back to skipping the inner steps when the slow nested step
finishes.
$ hg -y loop 7 --nested
\r (no-eol) (esc)
loop [ ] 0/7\r (no-eol) (esc)
loop [=====> ] 1/7\r (no-eol) (esc)
loop [============> ] 2/7\r (no-eol) (esc)
loop [===================> ] 3/7\r (no-eol) (esc)
loop [==========================> ] 4/7\r (no-eol) (esc)
nested [==========================> ] 3/5\r (no-eol) (esc)
nested [===================================> ] 4/5\r (no-eol) (esc)
loop [=================================> ] 5/7\r (no-eol) (esc)
loop [========================================> ] 6/7\r (no-eol) (esc)
\r (no-eol) (esc)
$ hg --config progress.changedelay=0 -y loop 3 --nested
\r (no-eol) (esc)
loop [ ] 0/3\r (no-eol) (esc)
nested [ ] 0/2\r (no-eol) (esc)
nested [======================> ] 1/2\r (no-eol) (esc)
loop [===============> ] 1/3\r (no-eol) (esc)
nested [ ] 0/2\r (no-eol) (esc)
nested [======================> ] 1/2\r (no-eol) (esc)
loop [===============================> ] 2/3\r (no-eol) (esc)
nested [ ] 0/2\r (no-eol) (esc)
nested [======================> ] 1/2\r (no-eol) (esc)
\r (no-eol) (esc)
test two topics being printed in parallel (as when we're doing a local
--pull clone, where you get the unbundle and bundle progress at the
same time):
$ hg loop 3 --parallel
\r (no-eol) (esc)
loop [ ] 0/3\r (no-eol) (esc)
loop [===============> ] 1/3\r (no-eol) (esc)
loop [===============================> ] 2/3\r (no-eol) (esc)
\r (no-eol) (esc)
test refresh is taken in account
$ hg -y --config progress.refresh=100 loop 3
test format options 1
$ hg -y --config 'progress.format=number topic item+2' loop 2
\r (no-eol) (esc)
0/2 loop lo\r (no-eol) (esc)
1/2 loop lo\r (no-eol) (esc)
\r (no-eol) (esc)
test format options 2
$ hg -y --config 'progress.format=number item-3 bar' loop 2
\r (no-eol) (esc)
0/2 p.0 [ ]\r (no-eol) (esc)
1/2 p.1 [=======================> ]\r (no-eol) (esc)
\r (no-eol) (esc)
test format options and indeterminate progress
$ hg -y --config 'progress.format=number item bar' loop -- -2
\r (no-eol) (esc)
0 loop.0 [ <=> ]\r (no-eol) (esc)
1 loop.1 [ <=> ]\r (no-eol) (esc)
\r (no-eol) (esc)
make sure things don't fall over if count > total
$ hg -y loop --total 4 6
\r (no-eol) (esc)
loop [ ] 0/4\r (no-eol) (esc)
loop [===========> ] 1/4\r (no-eol) (esc)
loop [=======================> ] 2/4\r (no-eol) (esc)
loop [===================================> ] 3/4\r (no-eol) (esc)
loop [===============================================>] 4/4\r (no-eol) (esc)
loop [ <=> ] 5/4\r (no-eol) (esc)
\r (no-eol) (esc)
test immediate progress completion
$ hg -y loop 0
test delay time estimates
#if no-chg
$ cp $HGRCPATH.orig $HGRCPATH
$ echo "[extensions]" >> $HGRCPATH
$ echo "mocktime=$TESTDIR/mocktime.py" >> $HGRCPATH
$ echo "progress=" >> $HGRCPATH
$ echo "loop=`pwd`/loop.py" >> $HGRCPATH
$ echo "[progress]" >> $HGRCPATH
$ echo "assume-tty=1" >> $HGRCPATH
$ echo "delay=25" >> $HGRCPATH
$ echo "width=60" >> $HGRCPATH
$ MOCKTIME=11 hg -y loop 8
\r (no-eol) (esc)
loop [=========> ] 2/8 1m07s\r (no-eol) (esc)
loop [===============> ] 3/8 56s\r (no-eol) (esc)
loop [=====================> ] 4/8 45s\r (no-eol) (esc)
loop [==========================> ] 5/8 34s\r (no-eol) (esc)
loop [================================> ] 6/8 23s\r (no-eol) (esc)
loop [=====================================> ] 7/8 12s\r (no-eol) (esc)
\r (no-eol) (esc)
$ MOCKTIME=10000 hg -y loop 4
\r (no-eol) (esc)
loop [ ] 0/4\r (no-eol) (esc)
loop [=========> ] 1/4 8h21m\r (no-eol) (esc)
loop [====================> ] 2/4 5h34m\r (no-eol) (esc)
loop [==============================> ] 3/4 2h47m\r (no-eol) (esc)
\r (no-eol) (esc)
$ MOCKTIME=1000000 hg -y loop 4
\r (no-eol) (esc)
loop [ ] 0/4\r (no-eol) (esc)
loop [=========> ] 1/4 5w00d\r (no-eol) (esc)
loop [====================> ] 2/4 3w03d\r (no-eol) (esc)
loop [=============================> ] 3/4 11d14h\r (no-eol) (esc)
\r (no-eol) (esc)
$ MOCKTIME=14000000 hg -y loop 4
\r (no-eol) (esc)
loop [ ] 0/4\r (no-eol) (esc)
loop [=========> ] 1/4 1y18w\r (no-eol) (esc)
loop [===================> ] 2/4 46w03d\r (no-eol) (esc)
loop [=============================> ] 3/4 23w02d\r (no-eol) (esc)
\r (no-eol) (esc)
Non-linear progress:
$ MOCKTIME='20 20 20 20 20 20 20 20 20 20 500 500 500 500 500 20 20 20 20 20' hg -y loop 20
\r (no-eol) (esc)
loop [=> ] 1/20 6m21s\r (no-eol) (esc)
loop [===> ] 2/20 6m01s\r (no-eol) (esc)
loop [=====> ] 3/20 5m41s\r (no-eol) (esc)
loop [=======> ] 4/20 5m21s\r (no-eol) (esc)
loop [=========> ] 5/20 5m01s\r (no-eol) (esc)
loop [===========> ] 6/20 4m41s\r (no-eol) (esc)
loop [=============> ] 7/20 4m21s\r (no-eol) (esc)
loop [===============> ] 8/20 4m01s\r (no-eol) (esc)
loop [================> ] 9/20 25m40s\r (no-eol) (esc)
loop [===================> ] 10/20 1h06m\r (no-eol) (esc)
loop [=====================> ] 11/20 1h13m\r (no-eol) (esc)
loop [=======================> ] 12/20 1h07m\r (no-eol) (esc)
loop [========================> ] 13/20 58m19s\r (no-eol) (esc)
loop [===========================> ] 14/20 7m09s\r (no-eol) (esc)
loop [=============================> ] 15/20 3m38s\r (no-eol) (esc)
loop [===============================> ] 16/20 2m15s\r (no-eol) (esc)
loop [=================================> ] 17/20 1m27s\r (no-eol) (esc)
loop [====================================> ] 18/20 52s\r (no-eol) (esc)
loop [======================================> ] 19/20 25s\r (no-eol) (esc)
\r (no-eol) (esc)
Time estimates should not fail when there's no end point:
$ MOCKTIME=11 hg -y loop -- -4
\r (no-eol) (esc)
loop [ <=> ] 2\r (no-eol) (esc)
loop [ <=> ] 3\r (no-eol) (esc)
\r (no-eol) (esc)
#endif
test line trimming by '[progress] width', when progress topic contains
multi-byte characters, of which length of byte sequence and columns in
display are different from each other.
$ cp $HGRCPATH.orig $HGRCPATH
$ cat >> $HGRCPATH <<EOF
> [extensions]
> progress=
> loop=`pwd`/loop.py
> [progress]
> assume-tty = 1
> delay = 0
> refresh = 0
> EOF
$ rm -f loop.pyc
$ cat >> loop.py <<EOF
> # use non-ascii characters as topic label of progress
> # 2 x 4 = 8 columns, but 3 x 4 = 12 bytes
> topiclabel = u'\u3042\u3044\u3046\u3048'.encode('utf-8')
> EOF
$ cat >> $HGRCPATH <<EOF
> [progress]
> format = topic number
> width= 12
> EOF
$ hg --encoding utf-8 -y loop --total 3 3
\r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88 0/3\r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88 1/3\r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88 2/3\r (no-eol) (esc)
\r (no-eol) (esc)
test calculation of bar width, when progress topic contains multi-byte
characters, of which length of byte sequence and columns in display
are different from each other.
$ cat >> $HGRCPATH <<EOF
> [progress]
> format = topic bar
> width= 21
> # progwidth should be 9 (= 21 - (8+1) - 3)
> EOF
$ hg --encoding utf-8 -y loop --total 3 3
\r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88 [ ]\r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88 [==> ]\r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88 [=====> ]\r (no-eol) (esc)
\r (no-eol) (esc)
test trimming progress items, when they contain multi-byte characters,
of which length of byte sequence and columns in display are different
from each other.
$ rm -f loop.pyc
$ rm -Rf __pycache__
$ cat >> loop.py <<EOF
> # use non-ascii characters as loop items of progress
> loopitems = [
> u'\u3042\u3044'.encode('utf-8'), # 2 x 2 = 4 columns
> u'\u3042\u3044\u3046'.encode('utf-8'), # 2 x 3 = 6 columns
> u'\u3042\u3044\u3046\u3048'.encode('utf-8'), # 2 x 4 = 8 columns
> ]
> def getloopitem(i):
> return loopitems[i % len(loopitems)]
> EOF
$ cat >> $HGRCPATH <<EOF
> [progress]
> # trim at tail side
> format = item+6
> EOF
$ hg --encoding utf-8 -y loop --total 3 3
\r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84 \r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\r (no-eol) (esc)
\r (no-eol) (esc)
$ cat >> $HGRCPATH <<EOF
> [progress]
> # trim at left side
> format = item-6
> EOF
$ hg --encoding utf-8 -y loop --total 3 3
\r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84 \r (no-eol) (esc)
\xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\r (no-eol) (esc)
\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88\r (no-eol) (esc)
\r (no-eol) (esc)