In MobileFrontend, we have code that provides the ability to load the
language info overlay without suggestions.
Currently /languages/all shows all languages without suggestions.
This patch will show suggestions on /languages/all but omit them
inside a new route: /languages/all/no-suggestions
Bug: T232410
Depends-On: Icce2631b2d5d80defc2287fb83f58f97a82a1b1e
Change-Id: Ide8e39b167f09842dd05eae7c26aeb741c29cd64
`lazy-image-placeholder` is needed only when MobileFrontend is enabled.
So it makes sense to move those styles to mobile.init css. See more details here: T199351#6380240
Bug: T260406
Change-Id: I16ca734af33fb0f3c4c67f20bbca8631279d7778
All images in MinervaNeue skin are vertically aligned `middle`,
so placeholder for them should be consistent.
Bug: T199351
Change-Id: I5fa00bd0c02e67059543389e4bbd07b073a2c5e0
This reverts commit ae8e3ad58b.
Reason for revert: This led to T260155 as the internal class uses
isSubPage which is not available on LinkTarget
Bug: T260155
Change-Id: I783f41e70cf1cf3b91737bce9cbf9ac729a778df
SkinUserPageHelper::__construct() should be able to consume a LinkTarget
too as NamespaceInfo::getSubjectPage() method returns a LinkTarget and
the Title::getSubjectPage() is deprecated.
Change-Id: I08bfc9bb21772b450798b6354bee2ab09d5dba4d
Hooks are global, hence any and all skin-specific OutputPageBodyAttributes hooks need to check first that the correct skin (or skins) is used, otherwise the class or classes get applied for *all* skins when only one skin is supposed to be targeted.
Also fixed a tiny code style issue in the same hook subscriber while at it.
Change-Id: I626b0e050c614687b31f1ffbaea92e371dee4574
I6e69b482cdbc adds a viewport with the same value as defined by Minerva.
Bug: T258290
Depends-On: I6e69b482cdbc49c09adc6f7cee54f5e17b7c92c4
Change-Id: Ie8ed80f734b20a16bc1c70a94d40290909180c3b
Disable sortable and collapsible during mediawiki.page.ready initialisation
Add onSkinPageReadyConfig hook that overrides sortable and collapsible values.
Adds 1.74kb of JS to all page loads.
Bug: T257265
Bug: T250851
Depends-On: I0dc38e74052027f26a70d58b5f520e5830e0d55d
Change-Id: I81ff55d080c8c4a40f425ca7d0fc4fda322055dd
OutputPage has similar method and they are being used
interchangeably here. Let's stick with one.
Bug: T257992
Change-Id: Ia5a0fbf114f6a1fec6fb85e01a408edae8a5badb
It was added in 7340485014
but for some reason never enabled.
This was supposed to be protecting us from performance regressions
by monitoring the size of modules in the repo. It appears to be
broken both locally and on Jenkins when run on the due to some changes
in Phabricator.
Since then we've added bundle size tests to Vector using a different
more foolproof method. I think that method is more preferable as it
also provides coverage for JS and icons so the
associated script is copied over and used in Minerva
and activated on all commits.
Bug: T259080
Change-Id: I9ba2dcc060ec09d91814c947d5be3f71b055a66e
I don't think it will fix the failing browser tests
but it will rule out Minerva / MobileFrontend as the cause in the
analysis.
This mirrors the codepath in Minerva that's triggering the warning.
Bug: T258096
Change-Id: Idd1c224c02e3d300889b6735fa1b9800212690ff
Previously they were just inheriting from `table`, but
now we don't require .infobox to use a specific element
they should be applied separately.
Fix table max-width override to use !important as this
too could be specified inline.
Bug: T258011
Change-Id: I4f28414fd965ae9f608b5e52047b9d474427853a
MoveLeadParagraphTransform treats div.infobox the same as table.infobox,
and should be fixed to match any element too.
Fix specificity of tablet overrides.
Bug: T258011
Change-Id: I9ada6d1b8b19b46dba6527f1f66edd2547fa0c88
