I'm going to assume the order in production with Chrome, as asserted
by the unit test, is the preferred order whereby negative numbers are
intended to be prepended before the positive numbers.
This came up in I6ecbd5743f3e, where this was the only unit test across
MediaWiki core and WMF gated extensions, that fails in Firefox.
It is reproducible locally with plain MW core + VisualEditor as well.
Bug: T366299
Change-Id: I6f8546e6e604cbb41e11bd2b59f8b5f19350c676
* Remove need for manual hacking of sub groups via "msg" strings
carefully prepended to every assertion.
* Improve CI details, by reporting the specific case that failed,
and local dev via ability to re-run each case, and reporting names
directly in the HTML Reporter and CLI summary.
* Reduce need for assert.async() and tracking of callbacks, especially
to improve failure details in case of Promise rejection.
Current logic was likely to cause a confusing timeout instead of a
clear failure if the promise ends up rejected.
QUnit propagates these as part of awaiting and asserting the test
closure's promise value (as async fn) automatically.
This approach also avoids the pitfal of a falsely passing test
when an assertion inside a done() handler was never reached.
* Use modern for-of where possible to remove need for closures and
arrow functions. Thus reducing complexity of test code, where
complexity should be kept lowest to avoid false confidence.
* Use plain for-in instead of overly complex Object.keys().forEach().
Change-Id: I934a266e75e64371081f104cfb867fb2c282c84a
This reverts commit a92dce4999.
A workaround for the previous problem was added in ContentTranslation
in I945897a27db479986855002b389034a745bf9bef.
Bug: T325249
Bug: T325566
Bug: T327779
Change-Id: I2d9c330dc4328468a65315ec6bed1d0f53ebd1f6
At first I was going for a more minimal replacement of mw.Uri with URL,
until I discovered that this code depends on a mw.Uri bug that would be
difficult to replicate:
// Expected: Relative URLs are accepted
new mw.Uri( '/foo' ).toString() // => 'https://localhost/foo'
// Expected: Protocol is optional
new mw.Uri( 'example.com/foo' ).toString() // => 'https://example.com/foo'
// Unexpected: Treated as empty domain with no protocol rather than relative URL
new mw.Uri( './foo' ).toString() // => 'https://./foo'
So I went for a bigger rewrite to preserve the intent rather than the
exact logic.
I had to change some test cases to use more realistic fake data. They
previously relied on bugs in our URL handling to pass despite the base
URLs being incorrect, particularly for non-short URLs (see T270219).
In my testing non-short URLs behave the same as before in practice.
Depends-On: I07a8c097dba0f5572c0aedf4febdf1434063ea6f
Bug: T325249
Change-Id: I232361266c1dda795b88018c3aaa3d9ecbe42b93
New changes:
2201b350c Localisation updates from https://translatewiki.net.
da74736c1 Remove unused test code
d1b016e90 Minor test tweaks
551de4f0c Specify document base URLs in more test cases
Local changes:
* Specify document base URLs in more test cases
Change-Id: I0e301bef38d97fa2234aa901c787360d9fbde8a3
Our encoding for the hrefs like "./Foo" that we send to Parsoid
differed slightly from how Parsoid outputs them, so to avoid dirty
diffs, we had to store the original ones we received from Parsoid
and send them back if they were unchanged.
Change the encoding to match Parsoid's exactly (by referring to the
Parsoid source code), and then remove 'rawTitle'/'origTitle'.
On a historical note, 'rawTitle'/'origTitle' were originally added to
fix other issues with links, which I hope are long behind us:
* bb45d984ca (T145978)
* fda2e6c1b5 (T44140)
Follow-up to 362df66b47, which removed
some other old stuff from the handling of Parsoid links.
Bug: T325766
Change-Id: I0ad0a655380eb2fb29b5ac01e2e399ac550ce34a
The dialog is unusable when there is no outline. See T313489 for a
longer explanation.
Bug: T313489
Change-Id: Ib2cc9c363d3596a16f6f1c4aef03ca216abf6b1f
Introducing a set method to have a different state for a set
parameter and a highlighted one in the selection.
Allows us to remove a lot of workarounds for missusing the
highlight state and fixes several issues with these workarounds.
Main implications:
- Keyboard navigation and mouse hover now sets the grey highlight
- If a parameter is set (blue highlight) keyboard navigation returns
when focusing the SelectWidget
- If nothing is set keyboard navigation starts at the top after focus
- Unchecking a parameter using space will not influence the keyboard
focus in the list
- Highlighting a parameter with the mouse lets keyboard navigation
continue from there.
Bug: T312647
Bug: T311204
Bug: T312213
Depends-On: I385dca1d95033961d3844e888521750443e49c95
Change-Id: Iaf089f4b271fd853b17c1aa7f5938510ea8f5431
Technically the old BookletLayout sidebar is still there. But it's
never visible, effectively dead, and meant to be removed. This patch
just empties the OutlineItems of all template dialog related pages
without actually disabling the old sidebar.
Note this partly reverts Ie57f462.
Bug: T310859
Bug: T310866
Change-Id: Ic0b7d703f369045ed342426563f8eeb3e47046db
This focuses on some scenarios that are
a) complex enough to be worth a test,
b) but simple enough so I don't need to spend hours on comming up
with a test setup. ;-)
This patch also simplifies the ARIA related code in
MWTransclusionOutlinePartWidget a bit.
* Check 1 of the 3 ARIA configs only. Only having one is already
helpful and should not be skipped.
* No need for the large conditional. setAriaDescribedBy() works fine
with undefined.
Bug: T291157
Change-Id: I142782ec9b96147de64497f4f6a373eae05b9c8e
The idea of the code under test is that it combines the results from
2 different API requests in a specific way. This is a first basic test
where both search results are small.
Bug: T291158
Change-Id: Ic4de57fb6b85afb952ea604769fddd06d44814c0
It removes some indirection.
* Find: \$\( '<.+?' \)\.toArray\(\)
* Replace with: $.parseHTML( \1 )
(also replaced a few by hand that weren't caught by that regexp)
Change-Id: I2639cc4a26bc960046a189504dd8058344e14d60
New changes:
cf7b2a9c4 Restore margin between wrapper paragraphs and document edge
629109b6b ve.ce.Surface tests: Sometimes assert originalDomElements in paste tests
3a5bd323c Use wrapper paragraphs in fixUpInsertion
Local changes:
* Update tests to ignore some internal properties
Bug: T299815
Bug: T299817
Change-Id: Ie52348f62320eb135c2f950ac78a77d9c665fef2
In a test case with 200 templates where all but a few parameters are
unused the loading time is cut in half.
Bug: T300974
Change-Id: Ice850cb9e5e95b9e3a19ff511b3a4f32117c7199
This focuses on a few trivial cases where the syntax helps making
the code more readable. One level of indirection is gone with this.
Change-Id: Ibf25d7eaa06952e69b36bd5a78a48d04ac62890c
These are only needed when we need to access a specific `this` from
within another `function () {}` context. This is not the case in the
situations here.
This is split from Ibf25d7e to make it smaller and easier to argue
about.
Change-Id: Ide1476de91fc343aa992ad92a1321d3a38b06dd0
New changes:
e5963f084 DiffElement: Remove doc-child-change wrappers
90a359f79 Follow-up Ia593dad8: Use modifyData when sanitizing
f5c7f0f20 Localisation updates from https://translatewiki.net.
Local changes:
* Update tests for the removal of doc-child-change wrappers
Change-Id: Ib29f23625de9151d50e0f3b7796b15ca9ed89133
The search field is of not much use when there is not really anything
to search. It wastes more space than the actual list of parameters.
Approved by UX, see T298259#7626538.
Bug: T298259
Change-Id: I01784a1c463d8b0b504897b20179719f91597d19
From the user's perspective this is the same as before: When a
template doesn't have any parameters, there is no search field. The
moment the first (undocumented) parameter is added the search field
appears.
This is just delayed now. The widgets are only created the moment
they are actually needed.
This saves loading time and memory, especially in a multi-part
transclusion with many zero-parameter templates.
This also makes it a lot easier to change the minimal number of
parameters from 1 to e.g. 4.
Includes reverting the flexible header composition done in
Ib050e30a50ef965c1524e977d3a600c3ff836774
Bug: T298259
Change-Id: Ied7541d8d5c0b478a439dd31ce072e634287f181
The "Add parameter" page always starts collapsed. Even if a template
doesn't contain anything but this. But most of the content isn't
visible, unless the user presses the button. It's not only a lot of
content, it's also rather expensive, including .parseDom(),
LinkCache.styleElement(), and ve.targetLinksToNewWindow(). This adds
up in large multi-part transclusions. In an example with 200 parts
the total blocking time goes down from 2.9s to 2.4s. Which means this
is not a major bottleneck, but still worth it.
Bug: T296335
Change-Id: Ieab9fd35d145142b04d2267d8e5a2e10a4c02784
This sorting algorithm was introduced via Ic6bc348 (T274544). Note
there is no index parameter in the .onReplacePart() handler at this
point. When a part was moved, it was removed and simply appended
to the end. The additional sorting was needed to move it back to the
correct position.
This changed a few days later via Iafe29f1. There is now an index
parameter. The .onReplacePart() handler does the same as before, but
puts the part at the correct position right away. The additional
sorting is pointless since then.
The removed code alone is responsible for 1/3 of the total blocking
time when the template dialog opens.
Bug: T296335
Change-Id: I6c3fa70b532d34cd29d59c3b48ab81ebf608d548
When changing the source in the described-by attributes the screen
reader will read the text of the new source when the status changes.
Just changing the text within the elements holding the descriptions
does not work.
Bug: T291284
Change-Id: I31cc3061cf6c1f699babe41e99e0711f0eb03646
