in favour of mw 1.34 shared.css to avoid weird inconsistent visual
styles
Inconsistencies include: monobook-style boxes, random text resizes,
miscellaneous inappropriate colours, unexpected centering of text
Use 'toc' even though it also has weird monobook box styles because
the less/js/rl/php soup for the collapsing is basically a labyrinth
and I can't replicate that; just override everything back to
basically how it's supposed to be, style-wise.
Incidentally resolves T239914 for timeless due to also overriding
toccolours styles that come with the toc feature.
Also include a fork of commonPrint.css as an accompaniment to
shared.css, which willl now hilariously go full-monobook. Ish.
TODO: Don't just use an entire copy of 1.34's shared.css and
commonPrint.css as a workaround, and come up with a better
approach to the toc maybe? (T280747)
Bug: T280285
Bug: T280292
Bug: T239914
Change-Id: I2d600a7d9768942b331f0de1dd1183e438144ad5
speshul junk
Assumes use of an svg icon because why would anyone seriously still
be using rasters in this day and age? Also probably negates the
usefulness of timeless' speshul logo handling junk anyway, as use of
correctly-sized svgs should obsolete the need for knowing the logo
size to begin with, as we no longer need to manually scale down
different source files...
Cap out at 156px wide, but other than that assume whatever size
people made it was intentional?
Change-Id: I9569e4d57c6740071e87c0f49ee47f739dd68bbd
Greenish is, uh, less shocking. I kept being confused. Other people
kept being confused. What in the world was I thinking? Why did this
of all things take me so long? How does one wake up? Take it from a
cat, it's probably not worth it...
Bug: T271632
Change-Id: I9b0225962e9103b8f0b33b6083dd564c47dbfb39
We need a smarter way to make things fit on pages with too many
cactions, and not remove useful labels for resolutions/pages
where they fit easily...
Bug: T271443
Change-Id: I3c696f18e2c4babf92c0af3aef85c6a0b72215fd
Let's just trust the users to not totally break mobile. They
absolutely will, but it's all on them now, because we're trusting
them, dammit.
But seriously, it's also a lot easier for them to fix this stuff
if they can see it to begin with...
Change-Id: I97c05aa35c1a6052870ff15b3900133c3ea16de5
Why the hork is this skin so damn complicated to begin with? A
saner DOM and it'd all just figure itself out, no problem...
Change-Id: I5409f3e1c14820332e94166f1ab24a343e20e8a1
Captions should stay put like regular content, as they're labels
for the table as a whole.
Acts kind of weird (jumpy) when the table is flowing off the page,
but totally stationary when whole table fits on page. Not sure
what this means.
Bug: T269912
Change-Id: I9df25315280aaae59dd06ec36f99832576c0c24c
No more overflow, but still scrollable. WHOO.
This is probably awful, but apparently I'm at least more capable
than I thought. That's probably not saying much.
Bug: T270050
Change-Id: Ieb7d2d756d4dad6ae8713edd2a1f19880420916b
* Put hover underline back on links, because wikimedians hate change
and also why not the entire point of the skin is to troll people,
let's do it right
* Adjust link colours for better distinction between hover, visited,
external
* Remove text labels from some obvious cactions (history,
addsection) so they're all more likely to fit on stupid pages
* Don't show empty languages toggle on mobile
* I'unno, catnip.
Bug: T181177
Bug: T181192
Bug: T265205
Change-Id: I6256d9e27d6019944987db62432b1b324d5c4b38
* Move images up to match helplink positioning, assuming standard
size of 20px
* Remove mobile header clear and let the float itself sort it out
if it even is long enough to matter (most are just icons)
Bug: T269550
Change-Id: Iaddab576b208f927c51c8f64b3e601fd89c0d710
* Add special case for multiselects for input padding
* More consistent styles for default forms
Bug: T259136
Change-Id: I1bea915a2887fdaf7e0bb3cc2ad17c26762688cf
YOU WANNA NOT MATCH THE CONTENT IN THE NAME OF MATCHING THE CONTENT,
FINE. I WON'T STOP YOU. FECKING AAAAARGH.
Because just feckin' matching the interface theme for top-level ones
would have made waaaaay too much sense... or, I dunno, taking the
average of *visible* backgrounds on the page to see if it's overall
light or dark...
But nope. Let's just use the body background, which is usually
entirely covered up anyway. How could that possibly go wrong?!
Change-Id: Ib31030815c947622aaad6340fbafbd6103fef21f
Also, like, maybe only trigger on top-level tables? Hopefully that
stops it from breaking everything... as much. Probably.
Bug: T269912
Bug: T269913
Change-Id: I09170d7dc67c49df4c010584ef0633a94659f29b
Avoid breaking legacy templates. What other templates were affected,
though? Was it just tables? Hopefully it is...
Bug: T181321
Bug: T269970
Change-Id: I2ce7d7b4344cc1b056feb99e47bd7f99959df454
* Make non-OOUI forms less fugly, and slightly more consistent,
without really bothering to actually do much of anything
* Make OOUI things line up a bit better
* Don't apply margin/padding to datetime widget (duh)
Change-Id: Iee4098b7fda2142c26356508b77dfe7c22d36dab
It's called shitty js. Affordance to label overflow is slightly wrong
for current behaviour, as it basically presents as a toggle. Maybe it
should toggle something! Like popouts, that would be fun. I'unno.
Slightly more framing on wikitables etc, but apply to all tables,
because why not.
Rejig to come with actual popouts or stuff...
Bug: T179265
Bug: T265785
Change-Id: I607cb87687fcbcb481dc2fb7165c52b9e9c35873
* Stop images from overflowing on desktop as well as mobile (folks wanna
overflow, they can load it sans skin, dammit)
* Firefox's stupid scrollbar colours
* Merge js into single module for desktop and mobile so it doesn't
randomly break on the wrong devices, and something about too many
modules anyway
* Weird footer padding?
* Weird WikiEditor/CodeMirror fonts/sizing
Bug: T256105
Bug: T247325
Change-Id: I8142c505534869c2530a915cfe8cef56303be6ed
This will only work with recent versions of core, but will not
give an error for older versions that don't have the
'mw-watchlink-temp' class.
Bug: T258674
Change-Id: Ia3d7732160dd0d8358ed12bd2c7ad08db16aa773
Apparently getFooterBlock never actually made it into core, so copied
out of Example.
Bug: T254266
Change-Id: Id62319314937080b99f2b93fd7b7fa62527d8fe7
Use normal `background-image` properties with SVGs now
that IE 8 and Android 2.1 is removed from Grade C.
Also removing all PNG fallback images.
Bug: T248062
Change-Id: I27d95241b6512f36985f7269122802d7b0962042
Monospace, serif, and sans-serif font stacks should all have consistent
metrics, so actually use them in wikitext editor. (Which depends on user
preferences.)
Also reset core font sizes added per T182320; that was done for skins
using system fonts elsewhere as well, and thus isn't really relevant here.
Also define out sans and serif font family variables independent of main
stack variables so they won't accidentally get overridden to other
things, but not presently following new naming conventions because
frankly I don't understand them and don't know which I should be using
to begin with. And even if I did, they're extremely long and we're
already exceeding maximum line length guidelines...
Bug: T247325
Change-Id: Iab98cfacab7dbc2f752bd200ee82d6d1a58c19a6
This feature was copied from Vector where it was recently removed
in e5c837e3ebf8.
In a nut shell, HTML output from PHP generally either has no
tabindex, tabindex=0 or tabindex=-1, or sometimes numbers between
1-100 for custom tab ordering. In JavaScript we then used
lastTabIndex (returns highest+1) to get an input field to be at the
end of the tab order. Thus making it be the *first* thing that
gets focussed when pressing Tab on a page (I guess the browser
offset is at the end of the document by default?)
This JS-based approach wasn't very stable as it required only
other thing in JS to use it for other reasons (e.g. genuinely
minting normal-use tab indexes for something) to break it as
we can't use the same logic for both "the lowest next number"
and "the highest possible number".
If we want to preserve this feature, it should probably do
something like attr('tabindex', '99999') which, while ugly,
would be a more sustainable way to implement this this.
Per Jack Phoenix in code review on this patch, match Vector
behaviour for now, where it was removed per T226148 for
being contrary to accessibility best practies.
Bug: T234581
Change-Id: I24777bdb6655f21d2f4896d5f887a125b36ab584
* Remove Liberation Mono from font stack; comes out excessively light on
hidpi chrome, to the point of unreadability (and probably is not a local
config problem beyond merely having it installed at all, as chrome
overrides normal font rendering settings regardless)
* Kill double border on pre styles; doesn't match other such boxes (notice,
warning, whatever)
Change-Id: I7a95931bd6af4efc880c3ee8f37a91d61f3e6539
jenkins-bot
Isarra
Perry Fraser
Sam Wilson
Volker E
Ed Sanders
Jack Phoenix
AronDemian
RetroCraft
Kunal Mehta
Timo Tijhof