bundle: emit full snapshot as is, without doing a redelta
With the new `forced` delta-reused policy, it become important to be able to
send full snapshot where full snapshot are needed. Otherwise, the fallback delta
will simply be used on the client side… creating monstrous delta chain, since
revision that are meant as a reset of delta-chain chain becoming too complex are
simply adding a new full delta-tree on the leaf of another one.
In the `non-forced` cases, client process full snapshot from the bundle
differently from deltas, so client will still try to convert the full snapshot
into a delta if possible. So this will no lead to pathological storage
explosion.
I have considered making this configurable, but the impact seems limited enough
that it does not seems to be worth it. Especially with the current
sparse-revlog format that use "delta-tree" with multiple level snapshots, full
snapshot are much less frequent and not that different from other intermediate
snapshot that we are already sending over the wire anyway.
CPU wise, this will help the bundling side a little as it will not need to
reconstruct revisions and compute deltas. The unbundling side might save a tiny
amount of CPU as it won't need to reconstruct the delta-base to reconstruct the
revision full text. This only slightly visible in some of the benchmarks. And
have no real impact on most of them.
### data-env-vars.name = pypy-2018-08-01-zstd-sparse-revlog
# benchmark.name = perf-bundle
# benchmark.variants.revs = last-40000
before: 11.467186 seconds
just-emit-full: 11.190576 seconds (-2.41%)
with-pull-force: 11.041091 seconds (-3.72%)
# benchmark.name = perf-unbundle
# benchmark.variants.revs = last-40000
before: 16.744862
just-emit-full:: 16.561036 seconds (-1.10%)
with-pull-force: 16.389344 seconds (-2.12%)
# benchmark.name = pull
# benchmark.variants.revs = last-40000
before: 26.870569
just-emit-full: 26.391188 seconds (-1.78%)
with-pull-force: 25.633184 seconds (-4.60%)
Space wise (so network-wise) the impact is fairly small. When taking compression into
account.
Below are tests the size of `hg bundle --all` for a handful of benchmark repositories
(with bzip, zstd compression and without it)
This show a small increase in the bundle size, but nothing really significant
except maybe for mozilla-try (+12%) that nobody really pulls large chunk of anyway.
Mozilla-try is also the repository that benefit the most for not having to
recompute deltas client size.
### mercurial:
bzip-before: 26 406 342 bytes
bzip-after: 26 691 543 bytes +1.08%
zstd-before: 27 918 645 bytes
zstd-after: 28 075 896 bytes +0.56%
none-before: 98 675 601 bytes
none-after: 100 411 237 bytes +1.76%
### pypy
bzip-before: 201 295 752 bytes
bzip-after: 209 780 282 bytes +4.21%
zstd-before: 202 974 795 bytes
zstd-after: 205 165 780 bytes +1.08%
none-before: 871 070 261 bytes
none-after: 993 595 057 bytes +14.07%
### netbeans
bzip-before: 601 314 330 bytes
bzip-after: 614 246 241 bytes +2.15%
zstd-before: 604 745 136 bytes
zstd-after: 615 497 705 bytes +1.78%
none-before: 3 338 238 571 bytes
none-after: 3 439 422 535 bytes +3.03%
### mozilla-central
bzip-before: 1 493 006 921 bytes
bzip-after: 1 549 650 570 bytes +3.79%
zstd-before: 1 481 910 102 bytes
zstd-after: 1 513 052 415 bytes +2.10%
none-before: 6 535 929 910 bytes
none-after: 7 010 191 342 bytes +7.26%
### mozilla-try
bzip-before: 6 583 425 999 bytes
bzip-after: 7 423 536 928 bytes +12.76%
zstd-before: 6 021 009 212 bytes
zstd-after: 6 674 922 420 bytes +10.86%
none-before: 22 954 739 558 bytes
none-after: 26 013 854 771 bytes +13.32%
==========================================================
Test various things around delta computation within revlog
==========================================================
basic setup
-----------
$ cat << EOF >> $HGRCPATH
> [debug]
> revlog.debug-delta=yes
> EOF
$ cat << EOF >> sha256line.py
> # a way to quickly produce file of significant size and poorly compressable content.
> import hashlib
> import sys
> for line in sys.stdin:
> print(hashlib.sha256(line.encode('utf8')).hexdigest())
> EOF
$ hg init base-repo
$ cd base-repo
create a "large" file
$ $TESTDIR/seq.py 1000 | $PYTHON $TESTTMP/sha256line.py > my-file.txt
$ hg add my-file.txt
$ hg commit -m initial-commit
DBG-DELTAS: FILELOG:my-file.txt: rev=0: delta-base=0 * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
Add more change at the end of the file
$ $TESTDIR/seq.py 1001 1200 | $PYTHON $TESTTMP/sha256line.py >> my-file.txt
$ hg commit -m "large-change"
DBG-DELTAS: FILELOG:my-file.txt: rev=1: delta-base=0 * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
Add small change at the start
$ hg up 'desc("initial-commit")' --quiet
$ mv my-file.txt foo
$ echo "small change at the start" > my-file.txt
$ cat foo >> my-file.txt
$ rm foo
$ hg commit -m "small-change"
DBG-DELTAS: FILELOG:my-file.txt: rev=2: delta-base=0 * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
created new head
$ hg log -r 'head()' -T '{node}\n' >> ../base-heads.nodes
$ hg log -r 'desc("initial-commit")' -T '{node}\n' >> ../initial.node
$ hg log -r 'desc("small-change")' -T '{node}\n' >> ../small.node
$ hg log -r 'desc("large-change")' -T '{node}\n' >> ../large.node
$ cd ..
Check delta find policy and result for merge on commit
======================================================
Check that delta of merge pick best of the two parents
------------------------------------------------------
As we check against both parents, the one with the largest change should
produce the smallest delta and be picked.
$ hg clone base-repo test-parents --quiet
$ hg -R test-parents update 'nodefromfile("small.node")' --quiet
$ hg -R test-parents merge 'nodefromfile("large.node")' --quiet
The delta base is the "large" revision as it produce a smaller delta.
$ hg -R test-parents commit -m "merge from small change"
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=1 * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
Check that the behavior tested above can we disabled
----------------------------------------------------
We disable the checking of both parent at the same time. The `small` change,
that produce a less optimal delta, should be picked first as it is "closer" to
the new commit.
$ hg clone base-repo test-no-parents --quiet
$ hg -R test-no-parents update 'nodefromfile("small.node")' --quiet
$ hg -R test-no-parents merge 'nodefromfile("large.node")' --quiet
The delta base is the "large" revision as it produce a smaller delta.
$ hg -R test-no-parents commit -m "merge from small change" \
> --config storage.revlog.optimize-delta-parent-choice=no
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
Check delta-find policy and result when unbundling
==================================================
Build a bundle with all delta built against p1
$ hg bundle -R test-parents --all --config devel.bundle.delta=p1 all-p1.hg
4 changesets found
Default policy of trusting delta from the bundle
------------------------------------------------
Keeping the `p1` delta used in the bundle is sub-optimal for storage, but
strusting in-bundle delta is faster to apply.
$ hg init bundle-default
$ hg -R bundle-default unbundle all-p1.hg --quiet
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=0: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=1: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=2: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob)
(confirm the file revision are in the same order, 2 should be smaller than 1)
$ hg -R bundle-default debugdata my-file.txt 2 | wc -l
\s*1001 (re)
$ hg -R bundle-default debugdata my-file.txt 1 | wc -l
\s*1200 (re)
explicitly enabled
------------------
Keeping the `p1` delta used in the bundle is sub-optimal for storage, but
strusting in-bundle delta is faster to apply.
$ hg init bundle-reuse-enabled
$ hg -R bundle-reuse-enabled unbundle all-p1.hg --quiet \
> --config storage.revlog.reuse-external-delta-parent=yes
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=0: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=1: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=2: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob)
(confirm the file revision are in the same order, 2 should be smaller than 1)
$ hg -R bundle-reuse-enabled debugdata my-file.txt 2 | wc -l
\s*1001 (re)
$ hg -R bundle-reuse-enabled debugdata my-file.txt 1 | wc -l
\s*1200 (re)
explicitly disabled
-------------------
Not reusing the delta-base from the parent means we the delta will be made
against the "best" parent. (so not the same as the previous two)
$ hg init bundle-reuse-disabled
$ hg -R bundle-reuse-disabled unbundle all-p1.hg --quiet \
> --config storage.revlog.reuse-external-delta-parent=no
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=0: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=1: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=2: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=1 * (glob)
(confirm the file revision are in the same order, 2 should be smaller than 1)
$ hg -R bundle-reuse-disabled debugdata my-file.txt 2 | wc -l
\s*1001 (re)
$ hg -R bundle-reuse-disabled debugdata my-file.txt 1 | wc -l
\s*1200 (re)
Check the path.*:delta-reuse-policy option
==========================================
Get a repository with the bad parent picked and a clone ready to pull the merge
$ cp -ar bundle-reuse-enabled peer-bad-delta
$ hg clone peer-bad-delta local-pre-pull --rev `cat large.node` --rev `cat small.node` --quiet
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=0: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=1: delta-base=0 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=2: delta-base=0 * (glob)
Check the parent order for the file
$ hg -R local-pre-pull debugdata my-file.txt 2 | wc -l
\s*1001 (re)
$ hg -R local-pre-pull debugdata my-file.txt 1 | wc -l
\s*1200 (re)
Pull with no value (so the default)
-----------------------------------
default is to reuse the (bad) delta
$ cp -ar local-pre-pull local-no-value
$ hg -R local-no-value pull --quiet
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob)
Pull with explicitly the default
--------------------------------
default is to reuse the (bad) delta
$ cp -ar local-pre-pull local-default
$ hg -R local-default pull --quiet --config 'paths.default:delta-reuse-policy=default'
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob)
Pull with no-reuse
------------------
We don't reuse the base, so we get a better delta
$ cp -ar local-pre-pull local-no-reuse
$ hg -R local-no-reuse pull --quiet --config 'paths.default:delta-reuse-policy=no-reuse'
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=1 * (glob)
Pull with try-base
------------------
We requested to use the (bad) delta
$ cp -ar local-pre-pull local-try-base
$ hg -R local-try-base pull --quiet --config 'paths.default:delta-reuse-policy=try-base'
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob)
Case where we force a "bad" delta to be applied
===============================================
We build a very different file content to force a full snapshot
$ cp -ar peer-bad-delta peer-bad-delta-with-full
$ cp -ar local-pre-pull local-pre-pull-full
$ echo '[paths]' >> local-pre-pull-full/.hg/hgrc
$ echo 'default=../peer-bad-delta-with-full' >> local-pre-pull-full/.hg/hgrc
$ hg -R peer-bad-delta-with-full update 'desc("merge")' --quiet
$ ($TESTDIR/seq.py 2000 2100; $TESTDIR/seq.py 500 510; $TESTDIR/seq.py 3000 3050) \
> | $PYTHON $TESTTMP/sha256line.py > peer-bad-delta-with-full/my-file.txt
$ hg -R peer-bad-delta-with-full commit -m 'trigger-full'
DBG-DELTAS: FILELOG:my-file.txt: rev=4: delta-base=4 * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
Check that "try-base" behavior challenge the delta
--------------------------------------------------
The bundling process creates a delta against the previous revision, however this
is an invalid chain for the client, so it is not considered and we do a full
snapshot again.
$ cp -ar local-pre-pull-full local-try-base-full
$ hg -R local-try-base-full pull --quiet \
> --config 'paths.default:delta-reuse-policy=try-base'
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=4: delta-base=4 * (glob)
Check that "forced" behavior do not challenge the delta, even if it is full.
---------------------------------------------------------------------------
A full bundle should be accepted as full bundle without recomputation
$ cp -ar local-pre-pull-full local-forced-full
$ hg -R local-forced-full pull --quiet \
> --config 'paths.default:delta-reuse-policy=forced'
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=4: delta-base=4 is-cached=1 - search-rounds=0 try-count=0 - delta-type=full snap-depth=0 - * (glob)
Check that "forced" behavior do not challenge the delta, even if it is bad.
---------------------------------------------------------------------------
The client does not challenge anything and applies the bizarre delta directly.
Note: If the bundling process becomes smarter, this test might no longer work
(as the server won't be sending "bad" deltas anymore) and might need something
more subtle to test this behavior.
$ hg bundle -R peer-bad-delta-with-full --all --config devel.bundle.delta=p1 all-p1.hg
5 changesets found
$ cp -ar local-pre-pull-full local-forced-full-p1
$ hg -R local-forced-full-p1 pull --quiet \
> --config 'paths.*:delta-reuse-policy=forced' all-p1.hg
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: CHANGELOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: MANIFESTLOG: * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob)
DBG-DELTAS: FILELOG:my-file.txt: rev=4: delta-base=3 * (glob)