phases: rework the logic of _pushdiscoveryphase to bound complexity
This rework the various graph traversal in _pushdiscoveryphase to keep the
complexity in check.
This is done though a couple of things:
- first, limiting the space we have to explore, for example, if we are not in
publishing push, we don't need to consider remote draft roots that are also
draft locally, as there is nothing to be moved there.
- avoid unbounded descendant computation, and use the faster "rev between"
computation.
This provide a massive boost to performance when exchanging with repository with
a massive amount of draft, like mozilla-try:
### data-env-vars.name = mozilla-try-2023-03-22-zstd-sparse-revlog
# benchmark.name = hg.command.push
# bin-env-vars.hg.flavor = default
# bin-env-vars.hg.py-re2-module = default
# benchmark.variants.explicit-rev = all-out-heads
# benchmark.variants.issue6528 = disabled
# benchmark.variants.protocol = ssh
# benchmark.variants.reuse-external-delta-parent = default
## benchmark.variants.revs = any-1-extra-rev
before: 20.346590 seconds
after: 11.232059 seconds (-38.15%, -7.48 seconds)
## benchmark.variants.revs = any-100-extra-rev
before: 24.752051 seconds
after: 15.367412 seconds (-37.91%, -9.38 seconds)
After this changes, the push operation is still quite too slow. Some of this
can be attributed to general phases slowness (reading all the roots from disk
for example) and other know slowness (not using persistent-nodemap, branchmap,
tags, etc. We are also working on them, but with this series, phase discovery
during push no longer showing up in profile and this is a pretty nice and bit
low-hanging fruit out of the way.
### (same case as the above)
# benchmark.variants.revs = any-1-extra-rev
pre-%ln-change: 44.235070
this-changeset: 11.232059 seconds (-74.61%, -33.00 seconds)
# benchmark.variants.revs = any-100-extra-rev
pre-%ln-change: 49.234697
this-changeset: 15.367412 seconds (-68.79%, -33.87 seconds)
Note that with this change, the `hg push` performance is now much closer to the
`hg pull` performance, even it still lagging behind a bit. (and the overall
performance are still too slow).
### data-env-vars.name = mozilla-try-2023-03-22-ds2-pnm
# benchmark.variants.explicit-rev = all-out-heads
# benchmark.variants.issue6528 = disabled
# benchmark.variants.protocol = ssh
# benchmark.variants.pulled-delta-reuse-policy = default
# bin-env-vars.hg.flavor = rust
## benchmark.variants.revs = any-1-extra-rev
hg.command.pull: 6.517450
hg.command.push: 11.219888
## benchmark.variants.revs = any-100-extra-rev
hg.command.pull: 10.160991
hg.command.push: 14.251107
### data-env-vars.name = mozilla-try-2023-03-22-zstd-sparse-revlog
# bin-env-vars.hg.py-re2-module = default
# benchmark.variants.explicit-rev = all-out-heads
# benchmark.variants.issue6528 = disabled
# benchmark.variants.protocol = ssh
# benchmark.variants.pulled-delta-reuse-policy = default
## bin-env-vars.hg.flavor = default
## benchmark.variants.revs = any-1-extra-rev
hg.command.pull: 8.577772
hg.command.push: 11.232059
## bin-env-vars.hg.flavor = default
## benchmark.variants.revs = any-100-extra-rev
hg.command.pull: 13.152976
hg.command.push: 15.367412
## bin-env-vars.hg.flavor = rust
## benchmark.variants.revs = any-1-extra-rev
hg.command.pull: 8.731982
hg.command.push: 11.178751
## bin-env-vars.hg.flavor = rust
## benchmark.variants.revs = any-100-extra-rev
hg.command.pull: 13.184236
hg.command.push: 15.620843
$ cat >> $HGRCPATH <<EOF
> [extensions]
> convert =
> [convert]
> hg.saverev = yes
> EOF
$ glog()
> {
> hg -R "$1" log -G --template '{rev} "{desc}" files: {files}\n'
> }
$ hg init source
$ cd source
$ echo a > a
$ echo b > b
$ echo f > f
$ hg ci -d '0 0' -qAm '0: add a b f'
$ echo c > c
$ hg move f d
$ hg ci -d '1 0' -qAm '1: add c, move f to d'
$ hg copy a e
$ echo b >> b
$ hg ci -d '2 0' -qAm '2: copy e from a, change b'
$ hg up -C 0
2 files updated, 0 files merged, 3 files removed, 0 files unresolved
$ echo a >> a
$ hg ci -d '3 0' -qAm '3: change a'
$ hg merge
merging a and e to e
3 files updated, 1 files merged, 1 files removed, 0 files unresolved
(branch merge, don't forget to commit)
$ hg ci -d '4 0' -qAm '4: merge 2 and 3'
$ echo a >> a
$ hg ci -d '5 0' -qAm '5: change a'
$ cd ..
Convert from null revision
$ hg convert --config convert.hg.startrev=null source full
initializing destination full repository
scanning source...
sorting...
converting...
5 0: add a b f
4 1: add c, move f to d
3 2: copy e from a, change b
2 3: change a
1 4: merge 2 and 3
0 5: change a
$ glog full
o 5 "5: change a" files: a
|
o 4 "4: merge 2 and 3" files: e
|\
| o 3 "3: change a" files: a
| |
o | 2 "2: copy e from a, change b" files: b e
| |
o | 1 "1: add c, move f to d" files: c d f
|/
o 0 "0: add a b f" files: a b f
$ rm -Rf full
Convert from zero revision
$ hg convert --config convert.hg.startrev=0 source full
initializing destination full repository
scanning source...
sorting...
converting...
5 0: add a b f
4 1: add c, move f to d
3 2: copy e from a, change b
2 3: change a
1 4: merge 2 and 3
0 5: change a
$ glog full
o 5 "5: change a" files: a
|
o 4 "4: merge 2 and 3" files: e
|\
| o 3 "3: change a" files: a
| |
o | 2 "2: copy e from a, change b" files: b e
| |
o | 1 "1: add c, move f to d" files: c d f
|/
o 0 "0: add a b f" files: a b f
Convert from merge parent
$ hg convert --config convert.hg.startrev=1 source conv1
initializing destination conv1 repository
scanning source...
sorting...
converting...
3 1: add c, move f to d
2 2: copy e from a, change b
1 4: merge 2 and 3
0 5: change a
$ glog conv1
o 3 "5: change a" files: a
|
o 2 "4: merge 2 and 3" files: a e
|
o 1 "2: copy e from a, change b" files: b e
|
o 0 "1: add c, move f to d" files: a b c d
$ cd conv1
$ hg up -q
Check copy preservation
$ hg st -C --change 2 e
M e
$ hg st -C --change 1 e
A e
a
$ hg st -C --change 0 a
A a
(It seems like a bug in log that the following doesn't show rev 1.)
$ hg log --follow --copies e
changeset: 2:8d3c3fe67bb7
user: test
date: Thu Jan 01 00:00:04 1970 +0000
summary: 4: merge 2 and 3
changeset: 0:23c3be426dce
user: test
date: Thu Jan 01 00:00:01 1970 +0000
summary: 1: add c, move f to d
Check copy removal on missing parent
$ hg log --follow --copies d
changeset: 0:23c3be426dce
user: test
date: Thu Jan 01 00:00:01 1970 +0000
summary: 1: add c, move f to d
$ hg cat -r tip a b
a
a
a
b
b
$ hg -q verify
$ cd ..
Convert from merge
$ hg convert --config convert.hg.startrev=4 source conv4
initializing destination conv4 repository
scanning source...
sorting...
converting...
1 4: merge 2 and 3
0 5: change a
$ glog conv4
o 1 "5: change a" files: a
|
o 0 "4: merge 2 and 3" files: a b c d e
$ cd conv4
$ hg up -C
5 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ hg cat -r tip a b
a
a
a
b
b
$ hg -q verify
$ cd ..
Convert from revset in convert.hg.revs
$ hg convert --config convert.hg.revs='3:4+0' source revsetrepo
initializing destination revsetrepo repository
scanning source...
sorting...
converting...
2 0: add a b f
1 3: change a
0 4: merge 2 and 3
$ glog revsetrepo
o 2 "4: merge 2 and 3" files: b c d e f
|
o 1 "3: change a" files: a
|
o 0 "0: add a b f" files: a b f
Convert from specified revs
$ hg convert --rev 3 --rev 2 source multiplerevs
initializing destination multiplerevs repository
scanning source...
sorting...
converting...
3 0: add a b f
2 1: add c, move f to d
1 2: copy e from a, change b
0 3: change a
$ glog multiplerevs
o 3 "3: change a" files: a
|
| o 2 "2: copy e from a, change b" files: b e
| |
| o 1 "1: add c, move f to d" files: c d f
|/
o 0 "0: add a b f" files: a b f
Convert in multiple steps that doesn't overlap - the link to the parent is
preserved anyway
$ hg convert --config convert.hg.revs=::1 source multistep
initializing destination multistep repository
scanning source...
sorting...
converting...
1 0: add a b f
0 1: add c, move f to d
$ hg convert --config convert.hg.revs=2 source multistep
scanning source...
sorting...
converting...
0 2: copy e from a, change b
$ glog multistep
o 2 "2: copy e from a, change b" files: b e
|
o 1 "1: add c, move f to d" files: c d f
|
o 0 "0: add a b f" files: a b f