Jun Wu <quark@fb.com> [Wed, 29 Mar 2017 14:49:14 -0700] rev 31765
verify: fix length check
According to the document added above, we should check L1 == L2, and the
only way to get L1 in all cases is to call "rawsize()", and the only way to
get L2 is to call "revision(raw=True)". Therefore the fix.
Meanwhile there are still a lot of things about flagprocessor broken in
revlog.py. Tests will be added after revlog.py gets fixed.
Jun Wu <quark@fb.com> [Wed, 29 Mar 2017 14:45:01 -0700] rev 31764
verify: document corner cases
It seems a good idea to list all kinds of "surprises" and expected behavior
to make the upcoming changes easier to understand.
Note: the comment added does not reflect the actual behavior of the current
code.
Denis Laxalde <denis.laxalde@logilab.fr> [Wed, 29 Mar 2017 22:26:16 +0200] rev 31763
hgweb: expose a followlines UI in filerevision view
In filerevision view (/file/<rev>/<fname>) we add some event listeners on
mouse clicks of <span> elements in the <pre class="sourcelines"> block.
Those listeners will capture a range of lines selected between two mouse
clicks and a box inviting to follow the history of selected lines will then
show up. Selected lines (i.e. the block of lines) get a CSS class which make
them highlighted. Selection can be cancelled (and restarted) by either
clicking on the cancel ("x") button in the invite box or clicking on any other
source line. Also clicking twice on the same line will abort the selection and
reset event listeners to restart the process.
As a first step, this action is only advertised by the "cursor: cell" CSS rule
on source lines elements as any other mechanisms would make the code
significantly more complicated. This might be improved later.
All JavaScript code lives in a new "linerangelog.js" file, sourced in
filerevision template (only in "paper" style for now).
Kostia Balytskyi <ikostia@fb.com> [Wed, 29 Mar 2017 05:31:31 -0700] rev 31762
shelve: move ui.quiet manipulations to configoverride
Jun Wu <quark@fb.com> [Thu, 30 Mar 2017 21:21:15 -0700] rev 31761
revlog: add a fast path for revision(raw=False)
If cache hit and flags are empty, no flag processor runs and "text" equals
to "rawtext". So we check flags, and return rawtext.
This resolves performance issue introduced by a previous patch.
Jun Wu <quark@fb.com> [Thu, 30 Mar 2017 18:38:03 -0700] rev 31760
revlog: make _addrevision only accept rawtext
All 3 users of _addrevision use raw:
- addrevision: passing rawtext to _addrevision
- addgroup: passing rawtext and raw=True to _addrevision
- clone: passing rawtext to _addrevision
There is no real user using _addrevision(raw=False). On the other hand,
_addrevision is low-level code dealing with raw revlog deltas and rawtexts.
It should not transform rawtext to non-raw text.
This patch removes the "raw" parameter from "_addrevision", and does some
rename and doc change to make it clear that "_addrevision" expects rawtext.
Archeology shows 2df983125d37 added "raw" flag to "_addrevision", follow-ups
e12c0fa1f65b and c1b7b2285522 seem to make the flag unnecessary.
test-revlog-raw.py no longer complains.
Jun Wu <quark@fb.com> [Thu, 30 Mar 2017 18:24:23 -0700] rev 31759
revlog: use raw revisions in clone
test-revlog-raw.py now shows "clone test passed", but there is more to fix.
Jun Wu <quark@fb.com> [Thu, 30 Mar 2017 18:23:27 -0700] rev 31758
revlog: use raw revisions in revdiff
See the added comment. revdiff is meant to output the raw delta that will be
written to revlog. It should use raw.
test-revlog-raw.py now shows "addgroupcopy test passed", but there is more
to fix.
Jun Wu <quark@fb.com> [Thu, 30 Mar 2017 17:58:03 -0700] rev 31757
revlog: use raw content when building delta
Using external content provided by flagprocessor when building revlog delta
is wrong, because deltas are applied to raw contents in revlog.
This patch fixes the above issue by adding "raw=True".
test-revlog-raw.py now shows "local test passed", but there is more to fix.
Jun Wu <quark@fb.com> [Thu, 30 Mar 2017 15:34:08 -0700] rev 31756
revlog: fix _cache usage in revision()
As documented at revlog.__init__, revlog._cache stores raw text.
The current read and write usage of "_cache" in revlog.revision lacks of
raw=True check.
This patch fixes that by adding check about raw, and storing rawtext
explicitly in _cache.
Note: it may slow down cache hit code path when raw=False and flags=0. That
performance issue will be fixed in a later patch.
test-revlog-raw now points us to a new problem.
Jun Wu <quark@fb.com> [Thu, 30 Mar 2017 14:56:09 -0700] rev 31755
revlog: rename some "text"s to "rawtext"
This makes code easier to understand. "_addrevision" is left untouched - it
will be changed in a later patch.
Jun Wu <quark@fb.com> [Thu, 30 Mar 2017 07:59:48 -0700] rev 31754
revlog: clarify flagprocessor documentation
The words "text", "newtext", "bool" could be confusing. Use explicit "text"
or "rawtext" and document more about the "bool".
Jun Wu <quark@fb.com> [Thu, 30 Mar 2017 20:48:57 -0700] rev 31753
revlog: add a stronger test for raw processing
There are some issues about revlog raw processing (flag processor). The test
is relatively strong covering many cases. It will verify fixes.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Fri, 31 Mar 2017 11:53:56 +0200] rev 31752
hook: add hook name information to external hook
While we are here, we can also add the hook name information to external hook.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Fri, 31 Mar 2017 11:08:11 +0200] rev 31751
hook: provide hook type information to external hook
The python hooks have access to the hook type information. There is not reason
for external hook to not be aware of it too.
For the record my use case is to make sure a hook script is configured for the
right type.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Fri, 31 Mar 2017 11:06:42 +0200] rev 31750
hook: use 'htype' in 'hook'
Same rational as for 'runhooks', we fix the naming in another function.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Fri, 31 Mar 2017 11:03:23 +0200] rev 31749
hook: use 'htype' in 'runhooks'
Same rational as for '_pythonhook', 'htype' is more accurate and less error
prone. We just fixed an error from the 'name'/'hname' confusion and this should
prevent them in the future.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Fri, 31 Mar 2017 11:02:05 +0200] rev 31748
hook: fix name used in untrusted message
The name used in the message we issue when a hook is untrusted was using "name"
which is actually the hook type and not the name of the hook.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Fri, 31 Mar 2017 10:59:37 +0200] rev 31747
hook: use "htype" as variable name in _pythonhook
We rename 'name' to 'htype' because it fits the variable content better.
Multiple python hooks already use 'htype' as a name for the argument. This makes
the difference with "hname" clearer and the code less error prone.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Thu, 30 Mar 2017 17:29:03 +0200] rev 31746
run-tests: auto-replace 'TXNID' output
Hooks related to the transaction are aware of the transaction id. By definition
this txn-id is unique and different for each transaction. As a result it can
never be predicted in test and always needs matching. As a result, touching any
like with this data is annoying. We solve the problem once and for all by
installing an automatic replacement. In test, this will now show as:
TXNID=TXN:$ID$
FUJIWARA Katsunori <foozy@lares.dti.ne.jp> [Sat, 01 Apr 2017 02:32:49 +0900] rev 31745
largefiles: use readasstandin() to read hex hash directly from filectx
BTW, C implementation of hexdigest() for SHA-1/256/512 returns hex
hash in lower case, and doctest in Python standard hashlib assumes
that, too. But it isn't explicitly described in API document or so.
Therefore, we can't assume that hexdigest() always returns hex hash in
lower case, for any hash algorithms, on any Python runtimes and
versions.
From point of view of that, it is reasonable for portability that
40800668e019 applies lower() on hex hash in overridefilemerge().
But on the other hand, in largefiles extension, there are still many
code paths comparing between hex hashes or storing hex hash into
standin file, without lower().
Switching to hash algorithm other than SHA-1 may be good chance to
clarify our policy about hexdigest()-ed hash value string.
- assume that hexdigest() always returns hex hash in lower case, or
- apply lower() on hex hash in appropriate layers to ensure
lower-case-ness of it for portability
FUJIWARA Katsunori <foozy@lares.dti.ne.jp> [Sat, 01 Apr 2017 02:32:49 +0900] rev 31744
largefiles: remove unused readstandin()
Now, there is no client of readstandin().
FUJIWARA Katsunori <foozy@lares.dti.ne.jp> [Sat, 01 Apr 2017 02:32:48 +0900] rev 31743
largefiles: make copytostore() accept only changectx as the 2nd argument (API)
As the name describes, the 2nd argument 'revorctx' of copytostore()
can accept non-changectx value, for historical reason,
But, since 10561eb97c7f, copyalltostore(), the only one copytostore()
client in Mercurial source tree, always passes changectx as
'revorctx'.
Therefore, it is reasonable to make copytostore() accept only
changectx as the 2nd argument, now.
FUJIWARA Katsunori <foozy@lares.dti.ne.jp> [Sat, 01 Apr 2017 02:32:48 +0900] rev 31742
largefiles: remove unused keyword argument of copytostore() (API)
AFAIK, 'uploaded' argument of copytostore() (or copytocache(), before
renaming at c65f5b6e26d4) has been never used both on caller and
callee sides, since official release of bundled largefiles extension.
FUJIWARA Katsunori <foozy@lares.dti.ne.jp> [Sat, 01 Apr 2017 02:32:48 +0900] rev 31741
largefiles: add copytostore() fstandin argument to replace readstandin() (API)
copyalltostore(), only one caller of copytostore(), already knows
standin file name of the target largefile. Therefore, passing it to
copytostore() is more efficient than calculating it in copytostore()
or readstandin().
FUJIWARA Katsunori <foozy@lares.dti.ne.jp> [Sat, 01 Apr 2017 02:32:47 +0900] rev 31740
largefiles: replace readstandin() by readasstandin()
These code paths already (or should, for efficiency at repetition)
know the target changectx and path of standin file.
FUJIWARA Katsunori <foozy@lares.dti.ne.jp> [Sat, 01 Apr 2017 02:32:31 +0900] rev 31739
largefiles: introduce readasstandin() to read hex hash from given filectx
This will be used to centralize and encapsulate the logic to read hash
from given (filectx of) standin file. readstandin() isn't suitable for
this purpose, because there are some code paths, which want to read
hex hash directly from filectx.
Ryan McElroy <rmcelroy@fb.com> [Thu, 30 Mar 2017 03:50:10 -0700] rev 31738
rebase: abort hg pull --rebase if rebase.requiredest is set (issue5514)
Previously, the pull would succeed, but the subsequent rebase would fail due
to the rebase.requiredest flag. Now abort earlier with a more useful error
message.
Ryan McElroy <rmcelroy@fb.com> [Thu, 30 Mar 2017 03:50:10 -0700] rev 31737
rebase: demonstrate behavior with requiredest and pull --rebase
Ryan McElroy <rmcelroy@fb.com> [Thu, 30 Mar 2017 03:50:10 -0700] rev 31736
rebase: allow destination-free continue and abort (issue5513)
Ryan McElroy <rmcelroy@fb.com> [Thu, 30 Mar 2017 03:50:10 -0700] rev 31735
rebase: test to show brokenness with requiredest
As shown in issue5513, --continue is broken when destination is required. This
adds a patch that demonstates this silly behavior, which will be fixed in a
future patch.
Ryan McElroy <rmcelroy@fb.com> [Thu, 30 Mar 2017 03:50:10 -0700] rev 31734
rebase: move destination test to new test file
We'll be adding a lot more tests here, so it makes sense to have this in its
own file now.
Matt Harbison <matt_harbison@yahoo.com> [Sat, 01 Apr 2017 00:21:17 -0400] rev 31733
templatefilter: add support for 'long' to json()
When disabling the '#requires serve' check in test-hgwebdir.t and running it on
Windows, several 500 errors popped up when querying '?style=json', with the
following in the error log:
File "...\\mercurial\\templater.py", line 393, in runfilter
"keyword '%s'") % (filt.func_name, dt))
Abort: template filter 'json' is not compatible with keyword 'lastchange'
The swallowed exception at that point was:
File "...\\mercurial\\templatefilters.py", line 242, in json
raise TypeError('cannot encode type %s' % obj.__class__.__name__)
TypeError: cannot encode type long
This corresponds to 'lastchange' being populated by hgweb.common.get_stat(),
which uses os.stat().st_mtime. os.stat_float_times() is being disabled in util,
so the type for the times is 'long' on Windows, and 'int' on Linux.
Denis Laxalde <denis@laxalde.org> [Thu, 30 Mar 2017 21:40:10 +0200] rev 31732
hgweb: prefix line id by ctx shortnode in filelog when patches are shown
When "patch" query parameter is present in requests to filelog view, line ids
in patches diff are no longer unique in the page since several patches are
shown on the same page. We now prefix line id by changeset shortnode when
several patches are displayed in the same page to have unique line ids
overall.
Jun Wu <quark@fb.com> [Wed, 29 Mar 2017 12:14:20 -0700] rev 31731
test-check-code: prevent files being added to the root directory
Adding new files in the root directory is probably a mistake, and is usually
discouraged [1]. The test catches it to avoid mistakes like [2].
Modify the test if files need to be added in the root.
[1]: https://www.mercurial-scm.org/pipermail/mercurial-devel/2016-July/086442.html
[2]: https://www.mercurial-scm.org/pipermail/mercurial-devel/2017-March/095836.html
Matt Harbison <matt_harbison@yahoo.com> [Wed, 29 Mar 2017 09:54:34 -0400] rev 31730
sslutil: clarify internal documentation
I ran into this python issue with an incomplete certificate chain on Windows
recently, and this is the clarification that came from that experimenting. The
comment I left on the bug tracker [1] with a reference to the CPython code [2]
indicates that the original problem I had is a different bug, but happened to
be mentioned under issue20916 on the Python bug tracker.
[1] https://bz.mercurial-scm.org/show_bug.cgi?id=5313#c7
[2] https://hg.python.org/cpython/file/v2.7.12/Modules/_ssl.c#l628
Jun Wu <quark@fb.com> [Wed, 29 Mar 2017 16:28:51 -0700] rev 31729
unionrepo: avoid unnecessary node -> rev conversion
Jun Wu <quark@fb.com> [Wed, 29 Mar 2017 16:28:00 -0700] rev 31728
bundlerepo: avoid unnecessary node -> rev conversion
Jun Wu <quark@fb.com> [Wed, 29 Mar 2017 16:23:04 -0700] rev 31727
revlog: avoid unnecessary node -> rev conversion
Jun Wu <quark@fb.com> [Wed, 29 Mar 2017 16:46:57 -0700] rev 31726
check-code: detect r.revision(r.node(rev))
revlog.revision takes either node or rev, but taking a rev is more
efficient, because converting rev to node is just a seek and read.
That's cheaper than converting node to rev, which may require O(n) walk in
revlog index for the first times, and then triggering building the radix
tree index. Even with the radix tree built, rev -> node is still faster than
node -> rev because the radix tree requires more jumps in memory.
So r.revision(r.node(rev)) should be changed to r.revision(rev). This patch
adds a check-code rule to detect that.
Jun Wu <quark@fb.com> [Wed, 29 Mar 2017 12:37:03 -0700] rev 31725
hardlink: check directory's st_dev when copying files
Previously, when copying a file, copyfiles will compare src's st_dev with
dirname(dst)'s st_dev, to decide whether to enable hardlink or not.
That could have issues on Linux's overlayfs, where stating directories could
result in different st_dev from st_dev of stating files, even if both the
directories and the files exist in the overlay's upperdir.
This patch fixes it by checking dirname(src) instead. It's more consistent
because we are checking directories for both src and dest.
That fixes test-hardlinks.t running on common Docker setups.
Jun Wu <quark@fb.com> [Wed, 29 Mar 2017 12:26:46 -0700] rev 31724
hardlink: duplicate hardlink detection for copying files and directories
A later patch will change one of them so they diverge.
Jun Wu <quark@fb.com> [Wed, 29 Mar 2017 12:21:15 -0700] rev 31723
hardlink: extract topic text logic of copyfiles
The topic text shows whether it's "linking" or "copying", based on
"hardlink" value. The function is extracted so a later patch can reuse it.
Pulkit Goyal <7895pulkit@gmail.com> [Wed, 29 Mar 2017 14:56:58 +0530] rev 31722
tests: add check to make sure summary works on Python 2
Pulkit Goyal <7895pulkit@gmail.com> [Wed, 29 Mar 2017 14:47:52 +0530] rev 31721
color: replace str() with pycompat.bytestr()
Pulkit Goyal <7895pulkit@gmail.com> [Sun, 26 Mar 2017 20:52:51 +0530] rev 31720
diff: slice over bytes to make sure conditions work normally
Both of this are part of generating `hg diff` on python 3.
Gregory Szorc <gregory.szorc@gmail.com> [Wed, 29 Mar 2017 20:19:26 -0700] rev 31719
minirst: remove redundant _admonitions set
As Yuya pointed out during a review a month ago, _admonitions and
_admonitiontitles are largely redundant. With the last commit, they
are exactly redundant. So, remove _admonitions and use
_admonitiontitles.keys() instead.
Gregory Szorc <gregory.szorc@gmail.com> [Wed, 29 Mar 2017 20:05:18 -0700] rev 31718
minirst: remove "admonition" from _admonitions
The "admonition" rst primitive is split into "specific" admonitions
("attention," "caution," etc) and the "generic" admonition
("admonition"). For more, see
http://docutils.sourceforge.net/docs/ref/rst/directives.html#admonitions
The _admonitions set and keys of the _admonitiontitles dict
overlap exactly except _admonitions has an "admonition" entry.
Nowhere in Mercurial is the "admonition" admonition directive used.
Even if it were, it doesn't have a title, so it wouldn't be rendered
correctly.
So, let's remove "admonition" from the set of recognized admonition
directives.
Gregory Szorc <gregory.szorc@gmail.com> [Wed, 29 Mar 2017 19:59:47 -0700] rev 31717
minirst: reindent _admonitiontitles
I don't like the verical indent.
While I was here, I cleaned up some whitespace and added a trailing
comma on the last element.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Tue, 28 Mar 2017 06:23:28 +0200] rev 31716
tags: extract filenode filtering into its own function
We'll also need to reuse this logic so we extract it into its own function. We
document some of the logic in the process.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Tue, 28 Mar 2017 06:08:12 +0200] rev 31715
tags: extract tags computation from fnodes into its own function
I'm about to introduce code that needs to perform such computation on
"arbitrary" nodes. The logic is extracted into its own function for reuse.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Tue, 28 Mar 2017 07:41:23 +0200] rev 31714
tags: only return 'alltags' in 'findglobaltags'
This is minor update along the way. We simplify the 'findglobaltags' function to
only return the tags. Since no existing data is reused, we know that all tags
returned are global and we can let the caller get that information if it cares
about it.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Tue, 28 Mar 2017 07:39:10 +0200] rev 31713
tags: make argument 'tagtype' optional in '_updatetags'
This is the next step from the previous changesets, we are now ready to use this
function in a simpler way.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Tue, 28 Mar 2017 07:38:10 +0200] rev 31712
tags: reorder argument of '_updatetags'
We move all arguments related to tagtype to the end, together. This will allow
us to make these arguments optional and reuse of this logic for callers that do
not care about the tag types.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Tue, 28 Mar 2017 06:13:49 +0200] rev 31711
tags: do not feed dictionaries to 'findglobaltags'
The code asserts that these dictionary are empty. So we can be more explicit
and have the function return the dictionaries directly.
Pierre-Yves David <pierre-yves.david@ens-lyon.org> [Tue, 28 Mar 2017 06:01:31 +0200] rev 31710
tags: extract fnode retrieval into its own function
My main goal here is to be able to reuse this logic easily. As a side effect
this important logic is now insulated and the code is clearer.
Denis Laxalde <denis.laxalde@logilab.fr> [Wed, 29 Mar 2017 12:07:07 +0200] rev 31709
hgweb: fix diff hunks filtering by line range in webutil.diffs()
The previous clause for filter out a diff hunk was too restrictive. We need to
consider the following cases (assuming linerange=(lb, ub) and the @s2,l2
hunkrange):
<-(s2)--------(s2+l2)->
<-(lb)---(ub)->
<-(lb)---(ub)->
<-(lb)---(ub)->
previously on the first and last situations were considered.
In test-hgweb-filelog.t, add a couple of lines at the beginning of file "b" so
that the line range we will follow does not start at the beginning of file.
This covers the change in aforementioned diff hunk filter clause.
Denis Laxalde <denis@laxalde.org> [Sat, 25 Mar 2017 11:30:08 +0100] rev 31708
summary: display obsolete state of parents
Extend the "parent: " lines in summary to display "(obsolete)" when the parent
is obsolete.
Denis Laxalde <denis@laxalde.org> [Sat, 25 Mar 2017 10:40:29 +0100] rev 31707
templates: add "changeset.obsolete" label in command line style
Following respective change in cmdutil.changeset_printer.
Denis Laxalde <denis@laxalde.org> [Tue, 28 Mar 2017 22:38:45 +0200] rev 31706
templates: shorten definition of changeset labels in command-line style
We'll add more labels and the line is already quite long, so let's define a
variable to hold all evolution "troubles" labels.
Denis Laxalde <denis@laxalde.org> [Tue, 28 Mar 2017 22:36:22 +0200] rev 31705
templates: use separate() to build changeset labels in command-line style
Denis Laxalde <denis@laxalde.org> [Sat, 25 Mar 2017 10:34:11 +0100] rev 31704
templatekw: add an "obsolete" keyword
Definition is the same as the one in evolve extension.
Denis Laxalde <denis@laxalde.org> [Sat, 25 Mar 2017 09:39:07 +0100] rev 31703
cmdutil: add a "changeset.obsolete" label in changeset_printer
Until now there were no label to highlight obsolete changesets in log output,
only evolution troubles (unstable, bumped, divergent) are supported. We add a
"changeset.obsolete" label on changeset entries produced by changeset_printer
so that obsolete changesets can be highlighted in log output. This is useful
because, unless using a graph log where obsolete changesets have a 'x' marker,
there's no way to identify obsolete changesets. And even in graph mode, when
working directory's parent is obsolete, we get a '@' marker and we do not see
it as obsolete.
Gregory Szorc <gregory.szorc@gmail.com> [Tue, 28 Mar 2017 14:40:13 -0700] rev 31702
fileset: perform membership test against set for status queries
Previously, fileset functions operating on status items performed
membership tests against a list of items. When there are thousands
of items having a specific status, that test can be extremely
slow. Changing the membership test to a set makes this operation
substantially faster.
On the mozilla-central repo:
$ hg files -r d14cac631ecc 'set:added()'
before: 28.120s
after: 0.860s
$ hg status --change d14cac631ecc --added
0.690s