The attrs package allows defining namedtuple-like classes with no weird
behavior and no runtime performance cost.
This patch vendors in attrs 17.2.0.
# no-check-commit
Differential Revision: https://phab.mercurial-scm.org/D867
--uncompressed isn't a very good name and its description in the
help documentation isn't very useful. We refer to this concept as
"stream clones" in a number of places. I think it makes sense to
change the user-facing argument to use the mode --stream. So this
commit does that.
We keep --uncompressed around for backwards compatibility.
While we're here, we overhaul the help docs for streaming clones
to be somewhat useful.
All tests have been updated to reflect the new preferred --stream
argument. A test for backwards compatibility of --uncompressed has
been added.
.. bc::
`hg clone --stream` should now be used instead of --uncompressed.
--uncompressed is marked as deprecated and is an alias for --stream.
There is no schedule for elimination of --uncompressed.
Differential Revision: https://phab.mercurial-scm.org/D864
This was added in 021454637716 in 2006. This may be possible but it
seems like a bad idea to even suggest it.
Differential Revision: https://phab.mercurial-scm.org/D863
Building on top of the new URL query string arguments to control
whitespace settings for annotate, this commit adds HTML checkboxes
reflecting the values of these arguments to the paper and gitweb
themes.
The actual diff settings are now exported to the templating layer.
The HTML templates add these as data-* attributes so they are
accessible to the DOM.
A new <form> with various <input> elements is added. The <form>
is initially hidden via CSS. A shared JavaScript function (which
runs after the <form> has been rendered but before the annotate
HTML (because annotate HTML could take a while to load and we want
the form to render quickly) takes care of setting the checked state
of each box from the data-* attributes. It also registers an event
handler to modify the URL and refresh the page whenever the checkbox
state is changed.
I'm using the URLSearchParams interface to perform URL manipulation.
https://developer.mozilla.org/en-US/docs/Web/API/URLSearchParams tells
me this may not be supported on older web browsers. Yes, apparently
the web API didn't have a standard API to parse and format query
strings until recently. Hence the check for the presence of this
feature in the JavaScript. If the browser doesn't support the
feature, the <form> will remain hidden and behavior will like it
currently is. We could polyfill this feature or implement our own
query string parsing. But I'm lazy and this could be done as a
follow-up if people miss it.
We could certainly expand this feature to support more diff options
(such as lines of context). That's why the potentially reusable code
is stored in a reusable place. It is also certainly possible to
add diff controls to other pages that display diffs. But since
Mozillians are making noise about controlling which revisions
annotate shows, I figured I'd start there.
.. feature::
Control whitespace settings for annotation on hgweb
/annotate URLs on hgweb now accept query string arguments to
influence how whitespace changes impact results.
The arguments "ignorews," "ignorewsamount," "ignorewseol," and
"ignoreblanklines" now have the same meaning as their [annotate]
config section counterparts. Any provided setting overrides the
server default.
HTML checkboxes have been added to the paper and gitweb themes
to expose current whitespace settings and to easily modify the
current view.
Differential Revision: https://phab.mercurial-scm.org/D850
This feature should hopefully be pretty straightforward. We simply
examine some query string arguments to feed into the diff options.
The function to obtain the diff options has been factored into its
own generic function to facilitate an upcoming change to the HTML
interface and to enable diff settings to be controlled via the same
query string arguments on other web commands.
The test output for "ignoreblanklines" didn't change. I'm not sure
why. Our test coverage for --ignore-blank-lines isn't great and I
can't figure out how to make this diff setting do anything meaningful.
On a very brief examination of the code, it is possible the setting
doesn't work because it is operating at the line level and blank lines
detection needs to examine multiple lines. But I'm not an expert in
this code, so I'm not sure.
Differential Revision: https://phab.mercurial-scm.org/D849
I will add another experiment in createmarkers, add a comment and some blank
lines for aesthetic sake.
Differential Revision: https://phab.mercurial-scm.org/D532
If the remote do not support phases, the "get" call will return None. We
change that default return to empty tuple to fix the membership testing.
This was a bug in the initial series. Thanks to yuja for catching this.
Previously, `hg clone --uncompressed` would always clone from the
origin server, even if a streaming clone bundle were available.
With this change, we invoke the clone bundle mechanism before the
stream clone mechanism, giving clone bundles the opportunity to
handle --uncompressed (which is mapped to pullop.streamclonepreferred).
The clone bundle filtering code now filters out entries that aren't
stream clones when a stream clone is requested. If a stream clone
clone bundle entry is present, it will be used. Otherwise, the client
will fall back to a server-based streaming clone.
.. feature::
`hg clone --uncompressed` uses clone bundles when possible
Differential Revision: https://phab.mercurial-scm.org/D833
Previously, chg's `verify` call could take 30+ms loading and checking new
config files. With one socket redirection, that adds up to around 70ms,
which is a lot for fast commands (ex. `bookmark --hidden`).
When investigating closer, A lot of time was spent on actually spent on ui
copying, which is mainly about `config.config` and `dict` copying.
This patch makes that 20x faster by adopting copy-on-write. The
copy-on-write is performed at config section level.
Before:
In [1]: %timeit ui.copy()
100 loops, best of 3: 2.32 ms per loop
After:
In [1]: %timeit ui.copy()
10000 loops, best of 3: 128 us per loop
2ms may look not that bad, but it adds up pretty quickly with multiple
calls. A typical chg run may call it 4 times, which is about 10ms.
Differential Revision: https://phab.mercurial-scm.org/D808
In this patch we returns False early if we are trying to check whether None is a
key in lazymanifest. The reason I added a diff is that on Python 3, it goes into
the lazy manifest code an returns a TypeError. I checked with Durham that
whether None can be a possible key in lazymanifest and he said "no". So it's
safe to have this if statement.
This fixes `hg merge` on Python 3.
Differential Revision: https://phab.mercurial-scm.org/D852
The arguments are especially non-obvious because the order is
different from dirstate.walk().
Differential Revision: https://phab.mercurial-scm.org/D847
The arguments are especially non-obvious because the order is
different from dirstate.status().
Differential Revision: https://phab.mercurial-scm.org/D846
This is a regression caused by 80d53a39fbcc. We have documented the behavior
in rebase help:
Rebase will destroy original commits unless you use "--keep". It will
also move your bookmarks (even if you do).
So let's restore the old behavior.
It is done by changing `scmutil.cleanupnodes` to accept more information so
a node could have different "movement destination" from "successors". It
also helps simplifying the callsite as a side effect - the special bookmark
movement logic in rebase is removed.
Differential Revision: https://phab.mercurial-scm.org/D727
The next patch will pass in overrides for bookmark moves, which is a
mapping itself, so let's rename "mapping" to "replacements" to
distinguish them better.
Differential Revision: https://phab.mercurial-scm.org/D749
We will soon want to pass in overrides for bookmark movements and this
will make that patch simpler. I also think this makes the code easier
to follow regardless of the later patch.
Differential Revision: https://phab.mercurial-scm.org/D748
A combination of wrapping `ui` and progress bars interrupted by exceptions can
lead to the progress bar not being cleared when the exception error is printed.
This results in corrupted-looking output like this:
```
updating [===============================> ] 1/2u
nresolved conflicts (see hg resolve, then hg rebase --continue)
```
This is because in `ui._progclear`, we only check the local reference to the
progress bar, not whether or not there is an instance of the singleton. When a
progress bar is interrupted by an exception, the exception printing in
`scmutil.callcatch` uses the original instance of the `ui` object, not the
wrapped copy that has `_progbar` set.
When consider whether or not to clear the progress bar, check for the existence
of the singleton, rather than just whether or not we have a local reference to
it.
Differential Revision: https://phab.mercurial-scm.org/D743
This is a trivial backport of 9f516a26a962 performed by
augie@google.com, but the change is still really Durham's not mine, so
I [augie] am leaving him as the author.
It seems like we used to pick the oldest possible version of the
changegroup to use for bundles created by the repair module (used
e.g. by "hg strip" and for temporary bundles by "hg rebase"). I tried
to preserve that behavior when I created the changegroup.safeversion()
method in 77f74106b264 (changegroup: introduce safeversion(),
2016-01-19).
However, we have recently chagned our minds and decided that these
commands are only used locally and downgrades are unlikely. That
decicion allowed us to start adding obsmarker and phase information to
these bundles. However, as the bug report shows, it means we get
different behavior e.g. when generaldelta is not enabled (because when
it was enabled, it forced us to use bundle2). The commit that actually
caused the reported bug was 26d535788092 (strip: include phases in
bundle (BC), 2017-06-15).
So, since we now depend on having more information in the bundles,
let's make sure we instead pick the newest possible changegroup
version.
Differential Revision: https://phab.mercurial-scm.org/D715
This was the original intent, but I bungled the logic. Otherwise if there is a
certificate chain issue, the repository can't be cloned in order for there to be
a repo object. I think I missed this case because I was inside of a Mercurial
clone as I was originally developing and testing this.
As part of moving dirstate storage to its own class, let's move the source of
truth for the dirstate parents to dirstatemap. This requires that dirstate._pl
no longer be a cache, and that all sets go through dirstatemap.setparents.
Differential Revision: https://phab.mercurial-scm.org/D759
As part of moving dirstate storage logic to a separate class, let's move the
function that reads the parents from the file. This will allow extensions to
write dirstate's that store the parents in other ways.
Differential Revision: https://phab.mercurial-scm.org/D758
As part of moving the dirstate storage logic to another class, let's move
opendirstatefile to dirstatemap. This will allow extensions to replace the
pending abstraction.
Future patches will move the consumers of _opendirstatefile into dirstatemap as
well.
Differential Revision: https://phab.mercurial-scm.org/D757