Yuya Nishihara <yuya@tcha.org> [Sun, 04 Oct 2015 22:25:29 +0900] rev 26492
util: extract stub function to get mtime with second accuracy
This function is trivial but will need a long comment why it can't use
st.st_mtime. See the next patch for details.
Siddharth Agarwal <sid0@fb.com> [Mon, 05 Oct 2015 14:37:59 -0700] rev 26491
scmutil: add a way for a repo's wlock to be inherited by a subprocess
This is part of a series that will allow locks to be inherited by subprocesses
in limited circumstances.
In the future, we'll call this for custom merge drivers.
Siddharth Agarwal <sid0@fb.com> [Mon, 05 Oct 2015 14:34:52 -0700] rev 26490
scmutil: add a way for a subprocess to be run with an inheritable lock
This is part of a series that will allow locks to be inherited by subprocesses
in limited circumstances.
In an upcoming patch, we'll add an API for the wlock to be inherited.
Siddharth Agarwal <sid0@fb.com> [Mon, 05 Oct 2015 14:27:37 -0700] rev 26489
localrepo: add a way to get the current wlock if it's held
This will be useful to pass around a reference to the lock to some functions
we're going to add to scmutil. We don't want those functions to live in
localrepo to avoid bloat.
Siddharth Agarwal <sid0@fb.com> [Mon, 05 Oct 2015 14:26:53 -0700] rev 26488
localrepo: add a way to get the current lock if it's held
We'll use this in upcoming patches to return a reference to the wlock if it's
held.
timeless@mozdev.org [Thu, 24 Sep 2015 13:58:18 -0400] rev 26487
test-help: verify that extension keywords appear in help templates
fixed by
a2291c9c85a1
Matt Harbison <matt_harbison@yahoo.com> [Mon, 24 Aug 2015 23:30:17 -0400] rev 26486
templatekw: add {changes}, {distance} and {tag} to the {latesttag} keyword
Matt Harbison <matt_harbison@yahoo.com> [Sun, 23 Aug 2015 23:44:58 -0400] rev 26485
templater: introduce {latesttag()} function to match a pattern (
issue4184)
This allows the latest class of tag to be found, such as a release candidate or
final build, instead of just the absolute latest.
It doesn't appear that the existing keyword can be given an optional argument.
There is a keyword, function and filter for 'date', so it doesn't seem harmful
to introduce a new function with the same name as an existing keyword. Most
functions are pretty Mercurial agnostic, but there is {revset()} as precedent.
Even though templatekw.getlatesttags() returns a single tuple, one entry of
which is a list, it is simplest to present this as a list of tags instead of a
single item, with each tag having a distance and change count attribute. It is
also closer to how {latesttag} returns a list of tags, and how this function
works when not given a '%' operator.
Matt Harbison <matt_harbison@yahoo.com> [Mon, 24 Aug 2015 23:07:00 -0400] rev 26484
templatekw: introduce showlatesttags() to handle {latesttag} keywords
The keywords {changes}, {distance} and {tag} will be available on a future
template method that will allow pattern matching against tag names. For
consistency, these should be available on the existing {latesttag} keyword as
well.
I debated whether or not to add {tag} instead of just continuing with the
existing {latesttag}. But it seems clearer not to have the same name for two
distinct things (a list in the LHS of %, and an individual tag value on the
right).
The value of latesttags[0] is the date of commit for the cset to which the tag
is applied (i.e. not the date the tag was applied), and therefore isn't made
visible because it doesn't seem interesting. It appears that this is merely an
internal implementation detail for sorting csets in a stable manner when there
are different branches.
Matt Harbison <matt_harbison@yahoo.com> [Mon, 05 Oct 2015 21:11:50 -0400] rev 26483
templatekw: factor out the changessincetag calculation to a private method
This will be reused in the next patch.
Matt Harbison <matt_harbison@yahoo.com> [Sun, 23 Aug 2015 23:22:55 -0400] rev 26482
templatekw: allow getlatesttags() to match a specific tag pattern
This will allow the latest class of tag to be found, such as a release candidate
or final build, instead of just the absolute latest. It will be exposed in a
future patch.
It's unfortunate that the original 'latesttags' cache can't be used to determine
the proper values, but it isn't fully populated for the entire repo. For
example, the {latesttagdistance} keyword on the Mecurial repo builds the cache
up back to the revision for 1.4. If the pattern was 're:^0\.\d$', that wouldn't
be in the cache. Maybe this can be optimized some other way, but for now, this
is the simpliest implementation.
Matt Harbison <matt_harbison@yahoo.com> [Sat, 22 Aug 2015 22:52:18 -0400] rev 26481
util: extract stringmatcher() from revset
This is used to match against tags, bookmarks, etc in revsets. It will be used
in a future patch to do the same tag matching in templater.
Gregory Szorc <gregory.szorc@gmail.com> [Mon, 05 Oct 2015 17:36:32 -0700] rev 26480
util.chunkbuffer: avoid extra mutations when reading partial chunks
Previously, a read(N) where N was less than the length of the first
available chunk would mutate the deque instance twice and allocate a new
str from the slice of the existing chunk. Profiling drawed my attention
to these as a potential hot spot during changegroup reading.
This patch makes the code more complicated in order to avoid the
aforementioned 3 operations.
On a pre-generated mozilla-central gzip bundle, this series has the
following impact on `hg unbundle` performance on my MacBook Pro:
before: 358.21 real 317.69 user 38.49 sys
after: 301.57 real 262.69 user 37.11 sys
delta: -56.64 real -55.00 user -1.38 sys
Gregory Szorc <gregory.szorc@gmail.com> [Mon, 05 Oct 2015 16:34:47 -0700] rev 26479
util.chunkbuffer: refactor chunk handling logic
This will make the next patch easier to read. It provides no benefit on
its own.
Gregory Szorc <gregory.szorc@gmail.com> [Mon, 05 Oct 2015 16:28:12 -0700] rev 26478
util.chunkbuffer: special case reading everything
The new code results in simpler logic within the while loop. It is also
faster since we avoid performing operations on the queue and buf
collections. However, there shouldn't be any super hot loops for this
since the whole point of chunkbuffer is to avoid reading large amounts
of data at once. This does, however, make it easier to optimize
chunkbuffer in a subsequent patch.
Mathias De Maré <mathias.demare@gmail.com> [Mon, 05 Oct 2015 07:13:35 +0200] rev 26477
revert: add reference to backout
Mathias De Maré <mathias.demare@gmail.com> [Mon, 05 Oct 2015 07:11:48 +0200] rev 26476
backout: add reference to revert
Siddharth Agarwal <sid0@fb.com> [Fri, 25 Sep 2015 12:39:23 -0700] rev 26475
localrepo: allow wlock to be inherited
This is part of a series that will allow locks to be inherited by subprocesses
in limited circumstances.
When allowed, the parent process will pass down requisite information to the
child process by way of this environment variable.
Siddharth Agarwal <sid0@fb.com> [Sun, 04 Oct 2015 20:04:44 -0700] rev 26474
lock.release: don't call postrelease functions for inherited locks
Review feedback from Pierre-Yves David. The postrelease functions typically
assume the lock is not held at all.
Siddharth Agarwal <sid0@fb.com> [Sun, 04 Oct 2015 20:02:50 -0700] rev 26473
lock: turn prepinherit/reacquire into a single context manager
Review feedback from Pierre-Yves David. This makes the overall code cleaner and
less error-prone, and makes a previously explicitly checked error state
impossible.
Siddharth Agarwal <sid0@fb.com> [Sun, 04 Oct 2015 19:28:43 -0700] rev 26472
localrepo: add a note about parentenvvar
Review feedback from Pierre-Yves David.
Gregory Szorc <gregory.szorc@gmail.com> [Sun, 04 Oct 2015 12:11:44 -0700] rev 26471
exchange: add "streaming all changes" to bundle2 pulling
This is the beginning of client-side support for performing a stream
clone using bundle2. The main bundle2 pull function checks whether to
perform a streaming clone and outputs a message if so.
While we have a duplicate message, it seems easier to have all the
bundle2 console writing in one location and in an easy-to-read
conditional block.
Gregory Szorc <gregory.szorc@gmail.com> [Sun, 04 Oct 2015 12:07:01 -0700] rev 26470
streamclone: move "streaming all changes" message location
Previously, the message was printed after we requested and started
processing the remote stream. This seems like something that we should
do before calling out to the remote. Moving it also makes it easier to
deal with the bundle2 implementation.
Gregory Szorc <gregory.szorc@gmail.com> [Sun, 04 Oct 2015 19:06:06 -0700] rev 26469
streamclone: move payload header generation into own function
The stream clone data over the wire protocol contains a header line
indicating total file count and data size. In bundle2, this metadata can
be captured by a part parameter and doesn't need to be in the body.
In preparation for bundle2, have generatev1() return the raw metadata
and move the header generation to its own function.