This allows for more accurate resolution of the width when compared
to the `resolve` method. This is more relevant for jQuery 1.x, where
the `resolve` method cannot find the width of a hidden select box,
but it also applies to newer versions of jQuery where the `width()`
method provided by jQuery doesn't fully match `getComputedStyle()`.
Fixes#3278Fixes#5502Closes#5259
* Add test for losing focus when searching tag entries
* Revert unknown unit test fix
Removing this no longer breaks a unit test, and having it in here
results in the select box receiving focus unexpectedly. It's not
clear what problem this was solving, since it was manually applied
from a series of pull requests.
It claims to be fixing an issue that was specific to IE11, and I'm
willing to re-introduce that bug because there doesn't appear to be
a regression test for it, and it's breaking some critical use cases.
The goal should be to focus the search box if it would have normally
lost focus when the selection was updated.
Fixes#5485Fixes#5516Closes#5550
* Update tests to be compatible with jQuery 3.0.0
There was a change in jQuery 3 that ensures that the return value of `.val()` on a multiple select is always an array. This is a breaking change from previous versions, where `null` or `undefined` were returned in these scenarios. Because we cannot `assert.equal` on a list of possible values, these assertions were switched to `assert.ok` which should be good enough.
* Properly strip out units in positioning tests
Before we were assuming that there were no units, and only were we stripping them out if we were expecting 3 digits. Now we just strip out all non-digit characters, so that should do the job and get us what we want.
There was a change in jQuery 3.2.0 that caused the units to be returned in these specific calls. They were not previously being returned, so this was not actually an issue.
* Add automated testing against jQuery 3.4.1
No tests appear to be currently failing.
We needed to define the `qunit` module in the unit tests because there was a change in grunt-contrib-qunit 0.6.0 that breaks when you define an AMD loader. It expects that the AMD loader is also used to load QUnit, instead of just being used to support the tests, so if you don't define the qunit module it will just hang and do nothing. Luckily we have the helpers file to help us out here, since it allows us to globally define this module.
* Start running tests against jQuery 2.x
We were only running tests against jQuery 1.x before and they were
all passing. This was a problem because apparently all of the data-*
attribute tests fail in jQuery 2.x. We are now running both the
integration and unit tests against both jQuery 1.x and jQuery 2.x.
Right now this resulted in a complete duplication of the test files
because there wasn't an obvious way to run the tests against both
versions. We're going to look into removing this duplication in the
future once the current issues are fixed.
We are also going to look into testing against jQuery 3.x in the
future, since that is also a supported line of jQuery.
* Force the data-* attributes to be parsed
There was a change made that switched us from using `$.data` and
`$.fn.data` internally to using an internal data store (managed
through internal utilities). This had the unfortunate side effect
of breaking the automatic loading of data-* options in versions of
jQuery other than 1.x, which included anything that would be
considered modern jQuery. While the change was made and approved
in good faith, all of the tests passed and the docs pages appeared
to be working, the tests really failed when running on newer versions
of jQuery. This was confirmed when we started running automated tests
against both versions, which confirmed the bug that others have been
seeing for a while.
The change was made becuase calling `$.fn.data` on an element which
contains a reference to itself in the internal jQuery data cache
would cause a stack overflow. This bug was well documented at the
following GitHub ticket and was resolved by no longer using
`$.fn.data`: https://github.com/select2/select2/issues/4014
Unfortunately because `$.fn.data` was no longer being called in a
way such that all of the data attributes would be dumped out, we
needed to find a replacement. The substitute that was given in the
original bug fix worked when the data cache was fully primed, but
we never primed it anywhere so it actually failed in the general
case. That meant we needed to find a way to manually prime it,
which is exactly what this change does.
* Clean up select2/utils
* Add scrollOnSelect as a configurable option
* default scrollOnSelect to true to avoid modifying existing behaviour
* added tests and default option for scrollAfterSelect
After we upgraded to QUnit 1.23.1, we gained support for
assert.expect(). This allows us to guard against any race conditions
within tests, because now expect() will be linked to the specific test
instead of the current running test.
This was required for us to get assert.async() support within tests, as
well as assert.expect() support. This was required because we need them
for multiple async tests that are coming.
These tests did not cover the classes that should have been
automatically applied to the dropdown based on the space around it. Now
they both test that the dropdown should be facing down, because there is
enough space below it to display the dropdown.
This adds a broken test that demonstrates the issue seen in
https://github.com/select2/select2/issues/3990 where existing selected
options are being reset once Select2 is initialized. This issue cannot
be reproduced on the options page [1] because the issue only appear to
happen if the selected option is not the first one in the list of
possible options.
[1]: https://select2.github.io/examples.html#data-array
This fixes the two failing assertions that only triggered in IE 9 (no
other versions) and Firefox. Both of them were caused by the offset for
the dropdown including a constant extra amount, what appeared to be
related to the size of the container if it actually had content. This
was not consistent in browsers, so now we are forcing there to be a
small amount of content within the container and then calculating the
expected offset based on that height.
There was a commit that landed in 4.0.1 that fixed positioning for
non-static elements, which are commonly used for the custom
`dropdownParent` option, but broke positioning for statically positioned
elements, commonly used in almost every other case. That commit was
c9216b4b96
This fixes the positioning issues caused by that commit by properly
calculating the offsets for statically positioned parents. Statically
positioned parents are unique, because the offset for the dropdown must
be calculated based on the closest element that is non-statically
positioned. Otherwise, the offset for any statically positioned parent
other than the body will be considerably higher than it should be,
resulting in the dropdown being offset by a large amount.
The offset parent for the body element is the html element, which is why
this works for both the body element and any custom parents for the
dropdown. This would not be needed if the parent wasn't customizable (as
seen in Select2 3.x) because you will never need to offset the body
element if it is statically positioned, because the html element almost
never has an offset.
This also fixes JSHint issues within the tests added in the last commit.
This closes https://github.com/select2/select2/issues/3970
This closes https://github.com/select2/select2/issues/3639
This adds a regression test that verifies the problem with positioning
the dropdown when the parent is a statically positioned element that
still has an offset. This could typically be seen if the body element
has an offset, which unfortunately it almost always does because of the
default user stylesheet in browsers. This was not caught during
pre-release testing because all of the test pages reset the margins and
padding on the body element.
This regression test verifies that the offsets that should be set for
the dropdown are calculated correctly. These were surprisingly difficult
to do because of how the offset is calculated using different
positioning techniques.
These tests are for https://github.com/select2/select2/issues/3970
These tests should fail because a selection can be removed even
though the container is disabled. This is because the only thing
preventing selections from being removed was the CSS which hid the
remove buttons when the container was disabled.
This adds the test that ensures that the search focus is still
focused, even after the selection is updated (for whatever reason).
Note that we are not triggering the `change` event here, and are
instead just re-calling `update` on the selection adapter. This is
because we do not bind the `change` event in tests, so the selection
is never re-rendered and the tests will pass. The `update` method
is triggered during the `change` cycle anyway, so this has the
same effect while supporting cases where the selection is re-rendered
without the selected values changing.
