Pierre-Yves David <pierre-yves.david@octobus.net> [Tue, 29 Nov 2022 19:54:55 +0100] rev 49683
peer-or-repo: make sure object in "scheme" have a `instance` object
The previous form of having heterogeneous object in the dictionnary makes things
more complicated than they needed to be. I am not super happy about the current
(especially around 'islocal', that most item do not have), but this is already
much better.
Pierre-Yves David <pierre-yves.david@octobus.net> [Tue, 29 Nov 2022 18:30:54 +0100] rev 49682
peer-or-repo: move the object setup in its own function
The `_peerorrepo` function is problematic, because it can build different types
of object (repository and peer). This make it hard to adjust the arguments to
the type of object we needs. So this patch start a series of change to create
peer and repo without going through a common function.
We move the part of the function doing object setup it its own function to make
it simpler to reuse in others contexts.
Pierre-Yves David <pierre-yves.david@octobus.net> [Sun, 06 Nov 2022 17:53:17 -0500] rev 49681
delta-find: use a single snapshot cache when applying a group to an object
This will avoid walking the revlog over and over again in some situations.
The difference is hard to show in our current benchmark suite, as the gain is
lower than their overall instability.
Pierre-Yves David <pierre-yves.david@octobus.net> [Sun, 06 Nov 2022 17:55:55 -0500] rev 49680
delta-find: make sure we only use newer full snapshot as candidate
The current code does not needs to protect against this, as there are no older
snapshot in the current cache. However as we are getting ready to reuse this
cache from one revision to another, we need the code to protect itself about
what's coming.
Pierre-Yves David <pierre-yves.david@octobus.net> [Sun, 06 Nov 2022 17:55:46 -0500] rev 49679
delta-find: use sets instead of list in the snapshot cache
This seems more appropriate.
Pierre-Yves David <pierre-yves.david@octobus.net> [Sun, 06 Nov 2022 16:56:23 -0500] rev 49678
delta-find: use a smarter object for snapshot caching
This open the way for a longer lived cache.
Pierre-Yves David <pierre-yves.david@octobus.net> [Mon, 07 Nov 2022 22:12:59 -0500] rev 49677
find-delta: pass the cache-delta usage policy alongside the cache-delta
The idea is to give higher level code more control to what will happens with
the cache delta passed. This should help with controling how we treat delta's
from different sources.
The final goal of this change is to allow for server modes where the client can
blindly accept any server delta without regards to any local constraints. This
will be implemented in later changesets.
Pierre-Yves David <pierre-yves.david@octobus.net> [Mon, 28 Nov 2022 18:58:35 +0100] rev 49676
find-delta: move most of the debug-find-delta code in the debug module
Lets us that module more. It will help us to keep revlog implementation details
close to each other.
Pierre-Yves David <pierre-yves.david@octobus.net> [Mon, 07 Nov 2022 20:02:32 -0500] rev 49675
find-delta: minor preparatory change
We are about to add more item in the cachedelta object, so lets access the item by index instead of doing a full extensions.
Pierre-Yves David <pierre-yves.david@octobus.net> [Mon, 07 Nov 2022 17:57:28 -0500] rev 49674
find-delta: rename _isgooddeltainfo
Lets move to a more readable name now that we are allowed to. This cannot hurt.
Pierre-Yves David <pierre-yves.david@octobus.net> [Mon, 07 Nov 2022 18:06:17 -0500] rev 49673
test-revlog-raw: drop the overwrite of dead code
The revlog class no longer have a _isgooddeltainfo method for a long time. So
overwriting it does not get us anything. The test have been wrapping the right
code since then anyway.
Pierre-Yves David <pierre-yves.david@octobus.net> [Sat, 12 Nov 2022 00:18:41 +0100] rev 49672
emitrevision: consider ancestors revision to emit as available base
This should make more delta base valid. This notably affects:
* case where we skipped some parent with empty delta to directly delta against
an ancestors
* case where an intermediate snapshots is stored.
This change means we could sent largish intermediate snapshots over the wire.
However this is actually a sub goal here. Sending snapshots over the wire means
the client have a high odd of simply storing the pre-computed delta instead of
doing a lengthy process that will… end up doing the same intermediate snapshot.
In addition the overall size of snapshot (or any level) is "only" some or the
overall delta size. (0.17% for my mercurial clone, 20% for my clone of Mozilla
try). So Sending them other the wire is unlikely to change large impact on the
bandwidth used.
If we decide that minimising the bandwidth is an explicit goal, we should
introduce new logic to filter-out snapshot as delta. The current code has no
notion explicite of snapshot so far, they just tended to fall into the wobbly
filtering options.
In some cases, this patch can yield large improvement to the bundling time:
### data-env-vars.name = mozilla-try-2019-02-18-zstd-sparse-revlog
# benchmark.name = perf-bundle
# benchmark.variants.revs = last-100000
before: 68.787066 seconds
after: 47.552677 seconds (-30.87%)
That translate to large improvement to the pull time :
### data-env-vars.name = mozilla-try-2019-02-18-zstd-sparse-revlog
# benchmark.name = pull
# benchmark.variants.
issue6528 = disabled
# benchmark.variants.revs = last-100000
before: 142.186625 seconds
after: 75.897745 seconds (-46.62%)
No significant negative impact have been observed.