phases: avoid N² behavior in `advanceboundary`
We allowed duplicated entries in the deque, which each entry could potentially
insert all its ancestors. So advancing boundary for the full repository would
mean each revision would walk all its ancestors, resulting in O(N²) iteration.
For repository of any decent size, N² is quickly insane.
We introduce a simple set to avoid this and get back to reasonable performance.
Check that the pull logger plugins logs pulls
=============================================
Enable the extension
$ echo "[extensions]" >> $HGRCPATH
$ echo "pull-logger = $TESTDIR/../contrib/pull_logger.py" >> $HGRCPATH
Check the format of the generated log entries, with a bunch of elements in the
common and heads set
$ hg init server
$ hg -R server debugbuilddag '.*2+2'
$ hg clone ssh://user@dummy/server client --rev 0
adding changesets
adding manifests
adding file changes
added 1 changesets with 0 changes to 0 files
new changesets 1ea73414a91b
updating to branch default
0 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ tail -1 server/.hg/pull_log.jsonl
{"common": ["0000000000000000000000000000000000000000"], "heads": ["1ea73414a91b0920940797d8fc6a11e447f8ea1e"], "logger_version": 0, "timestamp": *} (glob)
$ hg -R client pull --rev 1 --rev 2
pulling from ssh://user@dummy/server
searching for changes
adding changesets
adding manifests
adding file changes
added 2 changesets with 0 changes to 0 files (+1 heads)
new changesets d8736c3a2c84:fa28e81e283b
(run 'hg heads' to see heads, 'hg merge' to merge)
$ tail -1 server/.hg/pull_log.jsonl
{"common": ["1ea73414a91b0920940797d8fc6a11e447f8ea1e"], "heads": ["d8736c3a2c84ee759a2821385804bcb67f266ade", "fa28e81e283b3416de4d48ee0dd2d446e9e38d7c"], "logger_version": 0, "timestamp": *} (glob)
$ hg -R client pull --rev 2 --rev 3
pulling from ssh://user@dummy/server
searching for changes
adding changesets
adding manifests
adding file changes
added 1 changesets with 0 changes to 0 files
new changesets 944641ddcaef
(run 'hg update' to get a working copy)
$ tail -1 server/.hg/pull_log.jsonl
{"common": ["1ea73414a91b0920940797d8fc6a11e447f8ea1e", "fa28e81e283b3416de4d48ee0dd2d446e9e38d7c"], "heads": ["944641ddcaef174df7ce1bc2751a5f165129778b", "fa28e81e283b3416de4d48ee0dd2d446e9e38d7c"], "logger_version": 0, "timestamp": *} (glob)
Check the number of entries generated in the log when pulling from multiple
clients at the same time
$ rm -f server/.hg/pull_log.jsonl
$ for i in $($TESTDIR/seq.py 32); do
> hg clone ssh://user@dummy/server client_$i --rev 0
> done > /dev/null
$ for i in $($TESTDIR/seq.py 32); do
> hg -R client_$i pull --rev 1 &
> done > /dev/null
$ wait
$ wc -l server/.hg/pull_log.jsonl
\s*64 .* (re)
Test log rotation when reaching some size threshold
$ cat >> $HGRCPATH << EOF
> [pull-logger]
> rotate-size = 1kb
> EOF
$ rm -f server/.hg/pull_log.jsonl
$ for i in $($TESTDIR/seq.py 10); do
> hg -R client pull --rev 1
> done > /dev/null
$ wc -l server/.hg/pull_log.jsonl
\s*3 .* (re)
$ wc -l server/.hg/pull_log.jsonl.rotated
\s*7 .* (re)