tests/test-obsolete-tag-cache.t

changegroup: compute seen files as changesets are added (issue4750)

Before this patch, addchangegroup() would walk the changelog and compute
the set of seen files between applying changesets and applying
manifests. When cloning large repositories such as mozilla-central,
this consumed a non-trivial amount of time. On my MBP, this walk takes
~10s. On a dainty EC2 instance, this was measured to take ~125s! On the
latter machine, this delay was enough for the Mercurial server to
disconnect the client, thinking it had timed out, thus causing a clone
to abort.

This patch enables the changelog to compute the set of changed files as
new revisions are added. By doing so, we:

* avoid a potentially heavy computation between changelog and manifest
  processing by spreading the computation across all changelog additions
* avoid extra reads from the changelog by operating on the data as it is
  added

The downside of this is that the add revision callback does result in
extra I/O. Before, we would perform a flush (and subsequent read to
construct the full revision) when new delta chains were created. For
changelogs, this is typically every 2-4 revisions. Using the callback
guarantees there will be a flush after every added revision *and* an
open + read of the changelog to obtain the full revision in order to
read the added files. So, this increases the frequency of these
operations by the average chain length. In the future, the revlog
should be smart enough to know how to read revisions that haven't been
flushed yet, thus eliminating this extra I/O.

On my MBP, the total CPU times for an `hg unbundle` with a local
mozilla-central gzip bundle containing 251,934 changesets and 211,065
files did not have a statistically significant change with this patch,
holding steady around 360s. So, the increased revlog flushing did not
have an effect.

With this patch, there is no longer a visible pause between applying
changeset and manifest data. Before, it sure felt like Mercurial was
lethargic making this transition. Now, the transition is nearly
instantaneous, giving the impression that Mercurial is faster. Of course,
eliminating this pause means that the potential for network disconnect due
to channel inactivity during the changelog walk is eliminated as well.
And that is the impetus behind this change.

  $ cat >> $HGRCPATH << EOF
  > [extensions]
  > blackbox=
  > rebase=
  > mock=$TESTDIR/mockblackbox.py
  > 
  > [experimental]
  > evolution = createmarkers
  > EOF

Create a repo with some tags

  $ hg init repo
  $ cd repo
  $ echo initial > foo
  $ hg -q commit -A -m initial
  $ hg tag -m 'test tag' test1
  $ echo first > first
  $ hg -q commit -A -m first
  $ hg tag -m 'test2 tag' test2
  $ hg -q up -r 0
  $ echo newhead > newhead
  $ hg commit -A -m newhead
  adding newhead
  created new head
  $ hg tag -m 'test head 2 tag' head2

  $ hg log -G -T '{rev}:{node|short} {tags} {desc}\n'
  @  5:2942a772f72a tip test head 2 tag
  |
  o  4:042eb6bfcc49 head2 newhead
  |
  | o  3:c3cb30f2d2cd  test2 tag
  | |
  | o  2:d75775ffbc6b test2 first
  | |
  | o  1:5f97d42da03f  test tag
  |/
  o  0:55482a6fb4b1 test1 initial
  

Trigger tags cache population by doing something that accesses tags info

  $ hg tags
  tip                                5:2942a772f72a
  head2                              4:042eb6bfcc49
  test2                              2:d75775ffbc6b
  test1                              0:55482a6fb4b1

  $ cat .hg/cache/tags2-visible
  5 2942a772f72a444bef4bef13874d515f50fa27b6
  042eb6bfcc4909bad84a1cbf6eb1ddf0ab587d41 head2
  55482a6fb4b1881fa8f746fd52cf6f096bb21c89 test1
  d75775ffbc6bca1794d300f5571272879bd280da test2

Hiding a non-tip changeset should change filtered hash and cause tags recompute

  $ hg debugobsolete -d '0 0' c3cb30f2d2cd0aae008cc91a07876e3c5131fd22 -u dummyuser

  $ hg tags
  tip                                5:2942a772f72a
  head2                              4:042eb6bfcc49
  test1                              0:55482a6fb4b1

  $ cat .hg/cache/tags2-visible
  5 2942a772f72a444bef4bef13874d515f50fa27b6 f34fbc9a9769ba9eff5aff3d008a6b49f85c08b1
  042eb6bfcc4909bad84a1cbf6eb1ddf0ab587d41 head2
  55482a6fb4b1881fa8f746fd52cf6f096bb21c89 test1

  $ hg blackbox -l 4
  1970/01/01 00:00:00 bob> tags
  1970/01/01 00:00:00 bob> 2/2 cache hits/lookups in * seconds (glob)
  1970/01/01 00:00:00 bob> writing .hg/cache/tags2-visible with 2 tags
  1970/01/01 00:00:00 bob> tags exited 0 after * seconds (glob)

Hiding another changeset should cause the filtered hash to change

  $ hg debugobsolete -d '0 0' d75775ffbc6bca1794d300f5571272879bd280da -u dummyuser
  $ hg debugobsolete -d '0 0' 5f97d42da03fd56f3b228b03dfe48af5c0adf75b -u dummyuser

  $ hg tags
  tip                                5:2942a772f72a
  head2                              4:042eb6bfcc49

  $ cat .hg/cache/tags2-visible
  5 2942a772f72a444bef4bef13874d515f50fa27b6 2fce1eec33263d08a4d04293960fc73a555230e4
  042eb6bfcc4909bad84a1cbf6eb1ddf0ab587d41 head2

  $ hg blackbox -l 4
  1970/01/01 00:00:00 bob> tags
  1970/01/01 00:00:00 bob> 1/1 cache hits/lookups in * seconds (glob)
  1970/01/01 00:00:00 bob> writing .hg/cache/tags2-visible with 1 tags
  1970/01/01 00:00:00 bob> tags exited 0 after * seconds (glob)

Resolving tags on an unfiltered repo writes a separate tags cache

  $ hg --hidden tags
  tip                                5:2942a772f72a
  head2                              4:042eb6bfcc49
  test2                              2:d75775ffbc6b
  test1                              0:55482a6fb4b1

  $ cat .hg/cache/tags2
  5 2942a772f72a444bef4bef13874d515f50fa27b6
  042eb6bfcc4909bad84a1cbf6eb1ddf0ab587d41 head2
  55482a6fb4b1881fa8f746fd52cf6f096bb21c89 test1
  d75775ffbc6bca1794d300f5571272879bd280da test2

  $ hg blackbox -l 4
  1970/01/01 00:00:00 bob> --hidden tags
  1970/01/01 00:00:00 bob> 2/2 cache hits/lookups in * seconds (glob)
  1970/01/01 00:00:00 bob> writing .hg/cache/tags2 with 3 tags
  1970/01/01 00:00:00 bob> --hidden tags exited 0 after * seconds (glob)