| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
Change-Id: I34bf4e9b1c7ad7e95c0ff51935a8ff0986a240ac
|
|\ |
|
| |
| |
| |
| |
| | |
Bug: T325591
Change-Id: I52fe37b4c6d7b7d4ba7ccbc362413a996161156b
|
|\ \
| |/
|/| |
|
| |
| |
| |
| |
| |
| | |
Bug: T334855
Follows-Up: I278c54c682955c74bb6115a09e4a974c2b6e8ae6
Change-Id: I4b640c89618221c3892f4a9566a567ff1fbc3f7a
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
There are more other patches on SpecialRecentChanges but left
SpecialRecentChangesLinked untouched, some of them should also be
ported to this class.
But I wonder why these codes building the query are not in the
buildQuery() method, thus more of them can be reused.
Bug: T334242
Change-Id: I0a4f6e290271aa8de754daf0531be12f05aa4ca1
|
| | |
| | |
| | |
| | |
| | | |
Bug: T330641
Change-Id: Ie1772da59f92213fea598f02385c3e0ac6d200a7
|
|/ /
| |
| |
| |
| | |
Bug: T330641
Change-Id: I669e22b57f42bd7bd14ecf4511f15c04ff9f8558
|
| |
| |
| |
| |
| | |
Bug: T335025
Change-Id: I5a73a810d825422f1f2c08d502e5717ca7d06b21
|
| |
| |
| |
| | |
Change-Id: Id9de8549a0f9d016131a1c39d736e5304799d981
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This is upstreaming a pattern that's used in VisualEditor,
DiscussionTools, and (prompting this patch) MobileFrontend. These
extensions all take over certain views and want to avoid a flash of
non-taken-over content by using a hook to replace them with a "please
wait" loading message that'll redirect to a fallback page if the user
turns out to not have JavaScript enabled(or if it is but ResourceLoader
isn't available).
Change-Id: I8f25f53a05ce0437b0e9f63033cc151213c797bf
|
| |/
|/|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
When listing language links, we should include the dir attribute
for languages which do not match the direction. This makes the page
more machine-readable and allows for different stylistic treatment if
desired.
This will be helpful
for the mobile language overlay which needs to currently use the API
to look up language direction.
Change-Id: I54376d8314a75d115a11425db9c5b8696d9515d7
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
If an extension uses services from another extension, the previous
message was somewhat misleading: we only distinguish between “this
extension’s own services” and “anything else”. Make it a bit clearer
what the extension’s own services correspond to by including the prefix
in the message. (The other part, “MediaWiki services”, is still
inaccurate, but I can’t think of anything better.)
Change-Id: Ie509f813f1fd66572319ca66a2a13c802ef93fe3
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Since PHP 8.1, static variables in methods of a parent class are shared
between subclasses [1], so all ExtensionJsonTestBase subclasses were
actually seeing the same cached extension.json contents. Fix that by
explicitly moving the cache to a static class member (which is shared
between subclasses in all PHP versions) and splitting it by
$extensionJsonPath value.
[1]: https://www.php.net/manual/en/migration81.incompatible.php#migration81.incompatible.core.static-variable-inheritance
Change-Id: If5a0a8bb9f8b7c221169d57cfad2c5b9eadc9dd2
Co-Authored-By: Michael Große <michael.grosse@wikimedia.de>
|
| |
| |
| |
| | |
Change-Id: I31ca7f6db7730552f991b0a2e8b87c299e9d29e9
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
* Simplify async test by using async-await
Ref https://api.qunitjs.com/QUnit/test/.
* Use current coding style for separate var statements (instead of one-var),
and (within test suites) switch also to const/let.
* Remove needless QUnit.newMwEnvironment.
This module doesn't use mw.config or mw.msg, so there's no
need for QUnit.newMwEnvironment's hooks (pre-existing issue).
* Prefer QUnit.module scope over file scope.
Previously we often used file scope to ensure local variables
in classic debug=1 debug mode. Since QUnit 2 introduced module
scopes, useful for things like hooks.beforeEach(), we can use those
instead and not need an additional file closure regardless of which
debug mode is used.
* Prefer arrow functions for simple closures possible.
Showcased in upstream docs.
Ref https://api.qunitjs.com/QUnit/test/.
Bug: T250045
Change-Id: I4010200207de4d37701c9fb1e71fd962bdc40db9
|
|\| | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Analogous to what the 'compat/compat' rule indicates already, and the
fact that CI, fresh-node, and local dev generally involves recent
versions of Firefox, Chromium-based browsers, or Safari.
This patch is a minimal first step that only loosens the linter,
and auto-fixing only unusual cases where a new rule specifically
warned against existing code.
Change-Id: Ic64f650bd64669c57b884de3f75b23f3549b0c9f
|
|\ \ \ |
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Bug: T271030
Change-Id: I8ac6ebcb34c606acd1ecb4b160508bf265ef57c6
|
|\ \ \ \ |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Known titles should have blue links (showing existence) even if
they're reported as missing because they do not exist locally.
Bug: T333808
Change-Id: Iacb9dc04e532de27e41e938cac1af2f08401e949
|
|\ \ \ \ \ |
|
| | |/ / /
| |/| | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Gender support for the messages 'pageinfo-firstuser'
and 'pageinfo-lastuser'.
Bug: T334339
Change-Id: I0fb9f9e6e90266b1d58a5140c8fdb468c91d2729
|
|\ \ \ \ \
| |_|/ / /
|/| | | | |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The redirect targets another variant, but the target does not exist.
It causes infinite loop when querying the page with the redirect
and converttitles flags are both set.
Check the mConvertTitles to stop re-resolving the same title.
Bug: T333050
Change-Id: I64c5886dbd0a9148eeff033ba56306bcb7f2eb8c
|
|\ \ \ \ \ |
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
* TOCData in Parsoid expects to process non-string-key indexed arrays.
* Don't use 'null' as the default for maxtoclevel to ensure that
TOC is always displayed even when it isn't passed in as a param
by callers.
* Follows up on 05535be6 which only partially fixed the breakage
caused by 153a4157 and 439656e0
Bug: T334551
Change-Id: I8883b58574ea8ed0566de2c44dba3408a47d2d0c
|
|\ \ \ \ \ \ |
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
This reverts commit 0aa3d208a3cbd66f9c27f280abc529b4d641ee50 and
c71406a1ccf322a0a66f4b9551a96b2a4fc18013.
Reason for revert: Proper handling of complete input should be done
in the CheckUser extension instead of changing the input field type.
Bug: T334870
Change-Id: Ic5681b297ba2c4f3603c578d98b058639217645c
|
|\ \ \ \ \ \ \ |
|
| |/ / / / / /
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
Common naming convention for hook runner interfaces
Bug: T334689
Change-Id: I3043454a502289e5ee2492f7bf2ab01e155c30ba
|
|\ \ \ \ \ \ \ |
|
| |/ / / / / /
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
Common naming convention for hook runner interfaces
The hook is new in 1.40 and the rename should be backported,
but the old name needs alias as it is already used in extensions.
Bug: T334689
Change-Id: I3f9e6d33f1bf29e937c82860fcebd088dde647da
|
|\ \ \ \ \ \ \ |
|
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
Bug: T334391
Change-Id: I7e4965f369f76ba5ab55eaf8462068b719e3678c
|
|\ \ \ \ \ \ \ \ |
|
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
Replacing 'mediawiki.ui/variables.less' `@import` with
new skin-aware 'mediawiki.skin.variables.less' standard.
Also
- replacing several static values with new Codex design token featuring
skin variables of following properties:
- background color
- color
- border
- box shadow
- removing obsolete same-value variables from code base
- bringing Less imports in the same order, with variables first, as they
are used in mixins as well and always add `.less` extension to path
- renaming 'userlogin.css' to 'userlogin.less' to enable use of
skin variables consistently
Note,
- there's a small number of static values reintroduced where no
token is available yet. DS team will provide in follow-up patches
after a future Codex release.
- there are a number of color choices in the code, that should be
revisited by help of designers. They seem like non-standard choices.
Bug: T332541
Change-Id: I603a7bc52bf31233b2a82cdc2bff08fd1cb294fe
|
|\ \ \ \ \ \ \ \ \
| |/ / / / / / / /
|/| | | | | | | | |
|
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
This reverts commit 42aa5f9481433b5574fc4ff6bd4ea832c24adba0.
Reason for revert: Caused T334753, the proposed fix may need more time for review. Let's revert for now, before the train cut.
Bug: T310453
Bug: T334753
Change-Id: I790604eef00491b7f2a921fb3423a2f727f6593b
|
|\ \ \ \ \ \ \ \ \ |
|
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
When trying to purge thumbnails, currently it tries to purge from the
main backend only leading to unhappiness
Bug: T331138
Change-Id: I2b13912296df68b2833b8c7c0bfe17d4ba362fce
|
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
Change-Id: Ief701a709c5f100edad0ff0da04a4f3e0b1a8648
|
|\ \ \ \ \ \ \ \ \ \
| |_|_|_|/ / / / / /
|/| | | | | | | | | |
|
| | |_|/ / / / / /
| |/| | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
Bug: T334816
Change-Id: I46d835daf49eae113ce1a61f09e4a25432a25b20
|
|\ \ \ \ \ \ \ \ \
| |_|_|_|_|_|_|/ /
|/| | | | | | | | |
|
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
Rephrase this so as to reduce confusion about what is and isn't
supported. First and foremost these are primarly a reflection of
objective reality with regards to passing the checks, not an intent
of support. These are either supported Grade A, or browsers that are
equivalent enough and work under Grade X (e.g. less well-known
distros of the same engines, such as UC Browser, Vivaldi, Iceweasel;
or beta versions, and older versions that are new enough for our
purposes). They will always pass the checks whether we want them to
or not.
In extreme cases, we may improve unofficial support for a Grade X
browser by moving it explicitly into Grade C through a UA sniff, but
as of 6c7177349bd there are no longer such cases, and if we keep up
the pace [1], then it seems likely that there won't be new entries
or that UA sniffs are short-lived, staying only until the next bump.
I've removed the Grade C listing from here as it no longer has a
definition or purpose. Before WMF wikis required HTTPS in 2015, this
was effectively a duplicate of the wiki page with some added entries
for the browsers we UA-sniffed against. Since then, many older
browsers have been cut off from Basic support for reasons unrelated
to CSS or JS, and so it has largely gone out of date, presumably
on the idea that startup.js should list a superset of the "official"
page, just like we do for the Grade A version, except it's not clear
what a superset of Grade C support means, but whatever it is, its
bottom cut-off is not related by the cut-off for the JS pipeline,
rather it's related to purely in-writing what we want to support.
There is an infinite list of text browsers and unsupported browsers
that would in theory get Grade C if they were able to connect by
other means, but that's neither here nor there.
[1] https://wikitech.wikimedia.org/wiki/Performance/Guides/Frontend_performance_practices#General_approach
Bug: T178356
Change-Id: I8e403123205fae2e7b15b03b8c6c0cc8e9013ea9
|