Pulkit Goyal <pulkit@yandex-team.ru> [Mon, 10 Sep 2018 15:55:14 +0300] rev 39540
narrow: add narrow and ellipses as server capabilities
Right now we don't have a way to differentiate between whether a server can
serve ellipsis or not. The way we check whether a server is narrow enabled is by
checking bundle2 capability which does not scale outside of bundle2 world.
The goal is to use have wireprotocol commands just like remotefilelog for
widening the narrow clone, atleast in non-ellipses cases. Having a server
capability will help there as we can't rely on bundle2 capability there.
Also having a server capability is neat than having a bundle2 capability.
There are lot of things we can optimize locally on the client side by knowing
before hand that whether the server supports ellipses or not.
This will also help us in making sure that a client ellipses repo does not
communicate with a server repo without ellipses.
Differential Revision: https://phab.mercurial-scm.org/D4521
Kyle Lippincott <spectral@google.com> [Thu, 23 Aug 2018 18:26:14 -0700] rev 39539
treemanifest: use visitchildrenset when doing a walk
For this series, starting at 'introduce lazy loading of subdirs' and ending with
this commit, we get the following timing numbers, using roughly the same
methodology and setup that we did in an earlier commit.
"before" is 6268fed3.
diff --git:
repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
------+---+---+------------------------+-----------------------+------------
m-u | | | 1.585 s +- 0.034 s | 1.574 s +- 0.045 s | 99.3%
m-u | | x | 1.591 s +- 0.024 s | 1.601 s +- 0.034 s | 100.6%
m-u | x | | 1.579 s +- 0.032 s | 1.603 s +- 0.029 s | 101.5%
m-u | x | x | 109.8 ms +- 2.1 ms | 108.8 ms +- 2.2 ms | 99.1%
l-d-r | | | 234.6 ms +- 5.1 ms | 240.1 ms +- 7.9 ms | 102.3%
l-d-r | | x | 238.3 ms +- 8.1 ms | 232.2 ms +- 6.3 ms | 97.4%
l-d-r | x | | 120.0 ms +- 2.7 ms | 119.6 ms +- 2.1 ms | 99.7%
l-d-r | x | x | 120.2 ms +- 1.6 ms | 119.1 ms +- 1.6 ms | 99.1%
diff -c . --git:
repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
------+---+---+------------------------+-----------------------+------------
m-u | | | 356.9 ms +- 5.6 ms | 355.9 ms +- 5.6 ms | 99.7%
m-u | | x | 209.1 ms +- 3.3 ms | 205.8 ms +- 2.6 ms | 98.4%
m-u | x | | 418.3 ms +- 14.0 ms | 350.8 ms +- 7.7 ms | 83.9% <--
m-u | x | x | 168.3 ms +- 1.5 ms | 168.1 ms +- 2.7 ms | 99.9%
l-d-r | | | 100.5 ms +- 2.1 ms | 99.1 ms +- 1.6 ms | 98.6%
l-d-r | | x | 5.672 s +- 0.133 s | 4.335 s +- 0.051 s | 76.4% <--
l-d-r | x | | 102.8 ms +- 2.8 ms | 100.4 ms +- 1.7 ms | 97.7%
l-d-r | x | x | 1.025 s +- 0.028 s | 875.9 ms +- 12.1 ms | 85.5% <--
rebase -r . --keep -d .^^:
repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
------+---+---+------------------------+-----------------------+------------
m-u | | | 6.654 s +- 0.137 s | 6.718 s +- 0.046 s | 101.0%
m-u | | x | 6.672 s +- 0.104 s | 6.716 s +- 0.143 s | 100.7%
m-u | x | | 6.661 s +- 0.072 s | 6.658 s +- 0.066 s | 100.0%
m-u | x | x | 696.0 ms +- 49.0 ms | 696.2 ms +- 54.8 ms | 100.0%
l-d-r | | | 789.1 ms +- 12.5 ms | 793.5 ms +- 17.3 ms | 100.6%
l-d-r | | x | 7.621 s +- 0.194 s | 7.144 s +- 0.116 s | 93.7% <--
l-d-r | x | | 335.4 ms +- 9.7 ms | 339.6 ms +- 10.8 ms | 101.3%
l-d-r | x | x | 6.315 s +- 0.085 s | 5.633 s +- 0.105 s | 89.2% <--
status --change . --copies:
repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
------+---+---+------------------------+-----------------------+------------
m-u | | | 331.2 ms +- 7.0 ms | 335.1 ms +- 6.5 ms | 101.2%
m-u | | x | 184.0 ms +- 3.8 ms | 183.1 ms +- 3.4 ms | 99.5%
m-u | x | | 330.5 ms +- 7.2 ms | 332.7 ms +- 6.9 ms | 100.7%
m-u | x | x | 148.7 ms +- 3.5 ms | 147.5 ms +- 2.2 ms | 99.2%
l-d-r | | | 97.3 ms +- 1.3 ms | 98.2 ms +- 2.5 ms | 100.9%
l-d-r | | x | 5.196 s +- 0.030 s | 4.281 s +- 0.025 s | 82.4% <--
l-d-r | x | | 101.4 ms +- 3.1 ms | 100.9 ms +- 1.8 ms | 99.5%
l-d-r | x | x | 1.015 s +- 0.018 s | 861.2 ms +- 11.2 ms | 84.8% <--
status --copies:
repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
------+---+---+------------------------+-----------------------+------------
m-u | | | 2.356 s +- 0.032 s | 2.369 s +- 0.027 s | 100.6%
m-u | | x | 2.371 s +- 0.041 s | 2.385 s +- 0.028 s | 100.6%
m-u | x | | 2.366 s +- 0.036 s | 2.379 s +- 0.030 s | 100.5%
m-u | x | x | 121.8 ms +- 3.8 ms | 121.1 ms +- 2.7 ms | 99.4%
l-d-r | | | 726.2 ms +- 12.9 ms | 723.1 ms +- 12.7 ms | 99.6%
l-d-r | | x | 738.7 ms +- 11.2 ms | 741.3 ms +- 11.1 ms | 100.4%
l-d-r | x | | 215.9 ms +- 3.6 ms | 214.7 ms +- 5.9 ms | 99.4%
l-d-r | x | x | 213.1 ms +- 2.5 ms | 212.8 ms +- 4.5 ms | 99.9%
update $rev^; ~/src/hg/hg{hg}/hg update $rev:
repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
------+---+---+------------------------+-----------------------+------------
m-u | | | 3.922 s +- 0.045 s | 3.937 s +- 0.034 s | 100.4%
m-u | | x | 3.673 s +- 0.028 s | 3.660 s +- 0.035 s | 99.6%
m-u | x | | 3.959 s +- 0.073 s | 3.904 s +- 0.051 s | 98.6%
m-u | x | x | 417.0 ms +- 13.0 ms | 406.2 ms +- 6.5 ms | 97.4%
l-d-r | | | 534.3 ms +- 12.2 ms | 537.9 ms +- 8.8 ms | 100.7%
l-d-r | | x | 10.671 s +- 0.275 s | 10.111 s +- 0.114 s | 94.8% <--
l-d-r | x | | 309.7 ms +- 3.5 ms | 307.9 ms +- 4.0 ms | 99.4%
l-d-r | x | x | 1.837 s +- 0.014 s | 1.887 s +- 0.035 s | 102.7%
Differential Revision: https://phab.mercurial-scm.org/D4371
Kyle Lippincott <spectral@google.com> [Thu, 23 Aug 2018 00:44:10 -0700] rev 39538
treemanifest: use visitchildrenset when filtering a manifest to a matcher
Differential Revision: https://phab.mercurial-scm.org/D4370
Matt Harbison <matt_harbison@yahoo.com> [Mon, 10 Sep 2018 21:15:54 -0400] rev 39537
tests: stabilize test-no-symlink
This goes with 89630d0b3e23.
Boris Feld <boris.feld@octobus.net> [Tue, 29 May 2018 12:12:18 +0200] rev 39536
shelve: use the internal phase when possible
If the repository support it, use the internal phase for all changesets
created by shelve.
Kyle Lippincott <spectral@google.com> [Thu, 23 Aug 2018 00:41:20 -0700] rev 39535
treemanifest: avoid loading everything just to get their nodeid
Differential Revision: https://phab.mercurial-scm.org/D4369
Kyle Lippincott <spectral@google.com> [Thu, 23 Aug 2018 00:34:25 -0700] rev 39534
treemanifest: avoid unnecessary copies/processing when using alwaysmatcher
Differential Revision: https://phab.mercurial-scm.org/D4368