jQuery UI 1.13.0 changed the logic attaching the `_untrackClassesElement`
listener in the `_classes` widget method; one of the side effects was calling
`this._on` for each node that needed the listener. That caused a severe
performance degradation for large comboboxes as each `_on` jQuery UI call
causes a jQuery `add` call that calls Sizzle's `uniqueSort` underneath.
Instead, collect the nodes that need the listener and then, outside of the loop,
create a jQuery object out of them and attach the listener once. That's still
slower than the jQuery 1.12 version but only slightly: 936 ms to 1.03s on a very
large list on a recent MacBook Pro, compared to ~30 seconds before this patch.
Fixes gh-2014
Closes gh-2037
This helps fix issues that make tooltip tests sometimes fail when run against
jQuery 3.2 or newer due to timing differences.
Details:
* Add the `moduleAfterEach` function ensuring no timers are running.
* Attach this function via `common.testWidget`.
* Attach this function to most test suites.
* Add a tooltip test helper cleaning up leftover timers.
* Rename legacy `setup`/`teardown` hooks to `beforeEach`/`afterEach`.
Closes gh-1920
Summary of the changes:
* Build: Add jQuery 3.2.0-3.4.1 to versions UI can be tested against
* Build: Load jQuery & Migrate via HTTPS
* Build: Add package-lock.json to .gitignore
* Build: Update jQuery Migrate from 3.0.0 to 3.1.0
* Build: Allow to run tests against jQuery 3.x-git
* Build: Fix formatting according to JSCS rules
* Build: Disable JSCS for the inlined jQuery Color
* All: Switch from $.isArray to Array.isArray (jQuery.isArray will be
removed in jQuery 4.0)
* All: Switch from `$.isFunction( x )` to `typeof x === "function"`
(jQuery.isFunction will be removed in jQuery 4.0)
* All: Inline jQuery.isWindow as it'll be removed in jQuery 4.0
* Effects: Fix a timing issue in a variable declaration. Previously,
a jQuery object was created, chained & assigned to a variable that
was then accessed in a callback used inside of this chained
definition. Due to a timing difference in when the callback fired for
the first time in latest jQuery master, it was being called before
the variable was defined.
* Tests: Make dialog & draggable unit tests less strict (newest jQuery
returns fractional results in some cases, making comparisons fail when
there's a tiny difference)
* All: Migrate from $.trim to bare String.prototype.trim (jQuery.trim
will be deprecated in jQuery 3.5)
Closes gh-1901
`width: null` still matches the width of the original element.
`width: false` prevents an inline style from being set for the width. This
makes it easy to set the width via a stylesheet and allows the use of any
unit for setting the width, such as the new default of `14em`.
Fixes#11198
Closes gh-1467
So far we were using ui-state-active on active parent items,
ui-state-focus on active child items. The theme update highlighted the
visual inconsistency. With this change, only ui-state-active is used.
Fixes#10692
Selectmenu's test suite broke with b0e8380, which changed the padding
used by the menu widget. Selectmenu conditionally adds pixels to the
menu's width, and using closeEnough() accounts for that and the
rounding differences across browsers.
Closes gh-1275
