'repo' is a very confusing name to use for 'self', especially when
it's not a repo. Also drop repo.ui member (a.k.a. self.ui) now that
'self' doesn't shadow outer 'repo' variable.
This code is already used in a couple of places, and will need to be used in
others where wdir() support is needed. Trying to get the wdir() revision stored
in self._substate[1] when creating the object in subrepo.subrepo() resulted in
breaking various status and diff tests. This is a far less invasive change.
If the conversion doesn't change the hash, and the cset is public in the source,
it should be public in the destination. (This can happen if file remapping is
done that doesn't affect the initial commits.) This also propagates the secret
phase from the source, regardless of the hash, because presumably the content is
what is secret. Otherwise, the destination commit stays in the draft phase.
Maybe any draft cset with an unchanged hash should be changed to public, because
it has effectively been shared, but convert pretty strongly implies throwing
away (or at least readonly archiving) the source repo.
The change in the rollback output is because the name of the outer transaction
is now 'convert', which seems more accurate. Unfortunately, the memctx won't
indicate the hash prior to committing, so the proper phase can't be applied with
the commit.
The repo is already write locked in mercurial_sink.before().
This will be needed in the next patch to determine if phases need to be
adjusted. The insertion of the source revision in 'extras{}' is still
controlled by the config property.
Context: the result of computehidden, used to compute the 'visible' revisions
is cached. Its output can change when:
1) new obsolete commits are created
2) new bookmarks are created or deleted
3) new tags are created or deleted
4) the parents of the working copy change
We currently correctly invalidate the cache only in the case 1).
This patch fixes the second case (bookmarks) by invalidating the cache once
a bookmark is added or removed.
There were consistent test failures in test-bad-extension.t, because Windows
buffers stderr when redirected to a file (per the comment in ui.write_err()).
That resulted in failures like this:
--- c:/Users/Matt/Projects/hg/tests/test-bad-extension.t
+++ c:/Users/Matt/Projects/hg/tests/test-bad-extension.t.err
@@ -23,11 +23,11 @@
Traceback (most recent call last):
Exception: bit bucket overflow
*** failed to import extension badext2: No module named badext2
- Traceback (most recent call last):
- ImportError: No module named badext2
hg help [-ec] [TOPIC]
show help for a given topic or a help overview
+ Traceback (most recent call last):
+ ImportError: No module named badext2
show traceback for ImportError of hgext.name if debug is set
(note that --debug option isn't applied yet when loading extensions)
Instead of inserting another flush immediately after the print, to go along with
the one recently added prior to the print (see 3cd72de3be66), funnel the output
through ui.write_err(). The flush prior to printing the traceback only mentions
that stdout needs to be flushed, and only stderr needs to be flushed after
printing the traceback. ui.write_err() does both for us without needing to
redocument the quirky Windows behavior. It will also clear any progress bar.
As stated in the comment, using the smartset 'min' will give more opportunity to
be smart. It give a small but significant boost to the performance. Most of the
time is still spend doing the actual computation but at least we can scrap some
performance when it makes sense.
revset #0: roots(0:tip)
plain
0) 0.046600
1) 0.044109 94%
Python is slow at attributes lookup. No, really, I mean -slow-. prefetching
these three methods give use a measurable performance boost.
revset #0: 0::tip
plain
0) 0.037655
1) 0.034290 91%
This documentation was mostly intended for the user helps. However given the
lack of request for such feature, we should keep it un-documented. We stick the
help text in the code as it could still be useful to fellow contributors.
When we explicitly requested to update a bookmark but the bookmark location was
missing locally, we used to silently ignore the case. We now issue a message
about it to point that something wrong is going on.
By chance, we fixed all the cases where is case happened (for explicit pulling
only, issue4700 is still open). But I think it is still valuable to have a
warning in place in case such issue is reintroduced.
This patch have been tested against issue4689 test (but without issue4689 fix).
It give the better but expected failure seen below:
> --- /home/pyd/src/mercurial-dev/tests/test-bookmarks-pushpull.t
> +++ /home/pyd/src/mercurial-dev/tests/test-bookmarks-pushpull.t.err
> @@ -337,12 +337,12 @@
> adding manifests
> adding file changes
> added 1 changesets with 1 changes to 1 files
> - updating bookmark Y
> + remote bookmark Y point to locally missing 0d60821d2197
> (run 'hg update' to get a working copy)
> $ hg book
> * @ 1:0d2164f0ce0d
> X 1:0d2164f0ce0d
> - Y 5:35d1ef0a8d1b
> + Y 4:b0a5eff05604
> Z 1:0d2164f0ce0d
>
> Update a bookmark right after the initial lookup -r (issue4700)
> @@ -387,12 +387,11 @@
> adding manifests
> adding file changes
> added 1 changesets with 1 changes to 1 files
> - updating bookmark Y
> (run 'hg update' to get a working copy)
> $ hg book
> * @ 1:0d2164f0ce0d
> X 1:0d2164f0ce0d
> - Y 6:0d60821d2197
> + Y 4:b0a5eff05604
> Z 1:0d2164f0ce0d
> $ hg -R $TESTTMP/pull-race book
> @ 1:0d2164f0ce0d
The current fancyopts allows function as default parameter value
but not other callables.
By supporting other callables, we can have the benefits of e.g.,
custom __str__ method, which will be printed by 'hg help' as
the default value.
This was implied in issue3486, which specifically asked for subrepo support in
lfconvert. Now that lfconvert uses the convert extension internally when going
to normal files, the issue is half fixed. But now even non largefile repos
benefit when other transformations are needed.
Supporting a full subrepo tree conversion from a single command doesn't seem
reasonable, given the number of options that can be provided, and the
transformations that would need to occur when entering a subrepo (consider
'filemap' paths). Instead, this allows the user to incrementally convert each
hg subrepo from bottom up like so:
# so convert knows the dest type when it sees a non empty dest dir
$ hg init converted
$ hg convert orig/sub1 converted/sub1
$ hg convert orig/sub2 converted/sub2
$ hg convert orig converted
This allows different options to be applied to different subrepos more readily.
It assumes the shamap is in the default location in each converted subrepo for
simplicity. It also allows for a subrepo to be cloned into place, in case _it_
doesn't need a conversion. I was able to convert away from using
largefiles/bfiles in several subrepos with this mechanism.
Before this patch, when editing a hunk, if the editor was exiting with a non-
zero status the crecord session would be aborted. This patch makes crecord
handle the failure and goes back to the state before the failed edition.
We use the new error reporting mechanism introduced in the previous patch to
notify the user of the issue.
Before this patch, there was no way to report errors in the crecord ui.
This patch introduces a new variable errorstr. If set, its content replaces the
top banner displayed in crecord until a key is pressed.
This is useful to reuse the logic to implement the continuation of the crecord
session if the user edits a hunk and the editor exits with a non-zero status.
Sets have non-defined order and this should break stuff, but as we are lucky
fullreposet is also broken so the result is "not too bad".
We should fix it anyway, but it is too much for my current plate.
Using smartset's min will be significantly faster when the input set can provided
an optimised answer. I do not have time to fix all of them but I'm marking the
spot.
I cannot fix all issues in revset because I've got other things to do,
but let's write down all the brokenness to help other people reading
and fixing.
We remove revset making use of min and max as this is covered by the variants.
We could use variant for roots too, but it is not in the default so keep it
here.
We need more advanced variants in some cases. For example, "The last
rev of the sorted version".
We introduce a syntax for this: `reverse+last` means `last(reverse(REVSET))`.
We now use an 8 char display for timing (from 10), we add some logic to drop
precision if the number grows too large (as we do not care about sub-0 digit
in this case). This allow to pack more variants in a single screen.
The current benchmarks were only testing the whole iteration. This is suboptimal
because some changes are meaningful for things like first result, minimum or
sorting.
We introduce a "variants" feature that let you systematically add some variants
to all revsets tested.
A typical variants value would be 'plain,min,last,sort'. When testing 'all()' it
will also provide testing for:
- all()
- min(all())
- last(all())
- sort(sort)
and output:
plain min last sort
0) 0.034568 0.037857 0.000074 0.034238
1) 0.011358 32% 0.020181 53% 0.000080 108% 0.011405 33%
Using revsets (who hit the API) instead of the internal API add some overhead,
but the overhead should be the same everywhere so it still allow comparison.
This is is more simple to implement and allows comparison with older versions
who do not have the same API.
If the time difference is more than 5% from the previous run, we'll display
relative information. This makes it much simpler to spot performance changes in
a sea of benchmarks.
We mostly only care about total time. Dropping this output give us some room to
display more useful information (like percentage different) in future
changesets.
The file doc was saying something, the code was doing something else, the
argument validation was doing a third thing.
Doc and behavior now comply with the argument defined in the code.
We cannot just ask perfrevset to provide debug output because we usually want
to compare output from old version of Mercurial that do not support it. So, we
are using a regular expression.
(/we now have \d problems/).
For some reason (probably rebase issue, leprechaun or badly resolved .rej)
4d2ffbaf1d5d contains only half of the emailed patches and do not fix the bug.
This patch adds the other half and enable the sweet native computation for real.
As expected this provide massive speedup along the board.
revset #0: not public()
plain first
0) 0.011960 0.010523
1) 0.000465 3% 0.000492 4%
revset #1: (tip~1000::) - public()
plain first
0) 0.025700 0.025169
1) 0.002864 11% 0.001899 7%
revset #2: not public() and branch("default")
plain first
0) 0.022842 0.020863
1) 0.011418 49% 0.010948 52%
However, it has a less impact (even bad) on first result time in simple
situation. This comes from the overhead of building the set and filtering it.
This is especially true on my Mercurial repository (used here) where about 1/3
of the changesets are non public and hidden. This could be mitigated by a
caching of the set and a better usage of smartset in '_notpublic'. (But this
won't happen in this patch because the win is massive everywhere else).
revset #0: not public()
last
0) 0.000081
1) 0.000493 x6.1 <-- bad impact
revset #1: (tip~1000::) - public()
last
0) 0.013966
1) 0.002737 19%
revset #2: not public() and branch("default")
last
0) 0.011021
1) 0.011038
The effect mostly disappear when the number of non-public changesets is small
and/or the repo get bigger. Result for Mozilla central:
Mozilla
revset #0: not public()
plain first last
0) 0.092787 0.084094 0.000080
1) 0.000054 0% 0.000083 0% 0.000083
revset #1: (tip~1000::) - public()
plain first last
0) 0.215607 0.183996 0.124962
1) 0.031620 14% 0.006616 3% 0.031168 24%
revset #2: not public() and branch("default")
plain first last
0) 0.092626 0.082687 0.000162
1) 0.000139 0% 0.000165 0% 0.000167
Some pages, e.g. bookmarks, help and summary don't have a meaningful revision
context: they always either show information about tip or about the whole repo
(and not about any specific changeset). And error pages can just show hgweb
error messages, not related to any repo or changeset.
Having a hash in the links worked (even when '{node|short}' resolved to an
empty string on error pages), but seeing pages without revision context provide
links with hashes is a bit confusing (unless you keep current tip hash in your
head at all times) and not consistent with other template styles and other
links on the same page: they don't have a hash.
Let's just link to '/file', which is equal to '/file/tip'.
Some pages, e.g. bookmarks, help and summary don't have a meaningful revision
context: they always either show information about tip or about the whole repo
(and not about any specific changeset). And error pages can just show hgweb
error messages, not related to any repo or changeset.
When monoblue style was added in f9c487618909, however, all graph links had
tried to point at some hash, and on such pages as described above it didn't
make sense. On error pages '{node|short}' is empty string anyway.
Of course, it worked, but seeing such pages without revision context provide
links with hashes is a bit confusing (unless you keep current tip hash in your
head at all times) and wasn't consistent with other template styles, other
pages in monoblue and even other links on the same page.
Let's just link to '/graph', which is equal to '/graph/tip'.
