There is a lot of logic in intro.js; now we test four cases:
1. (implicitly, via QUnit tests) A real browser with window being the global
2. Browserify where there are both global & window variables.
3. Node with jsdom where window is passed manually to the jQuery factory.
4. Pure Node with incorrect window passed; jQuery should throw then.
Previously the second & fourth case was not tested and the third was tested
in a way that interfered with the main test environment.
We now also test if in the Browserify case we're not creating a jQuery global
by default.
Fixes gh-2181
Closes gh-2234
Android 2.3 is very slow & times out a lot in async tests, they have to be
restarted multiple times to settle. Long test execution is not a huge problem
as Android 2.3 is tested only periodically during the night, unstable tests
are a bigger problem. This might mitigate that.
In a regular scenario almost all tests should pass so increasing the timeout
for all browsers shouldn't have a huge impact on overall test time.
Closes gh-2232
jQuery.swap was an undocumented API used only internally. With the modular
AMD system we currently have it's not necessary to expose this function
publicly under the jQuery object.
Fixes gh-2058
Closes gh-2182
Most people just read the README that's displayed to them on GitHub which
is the one for the master branch. Let's include the info about Sizzle
not being excludable there.
Fixes gh-2184
The latest version supporting Node.js is 3.1.2; some workarounds are not needed
for this version. For example, in jsdom 3.1.2 a document created via
document.implementation.createHTMLDocument( "" ) has a body.
Fixes gh-2153
Closes gh-2154
BrowserStack regularly has outdated Java in their IE instances causing our
data tests to hang & give no output. Skip the tests that create applets,
applets are treated in the same way as an embed elements which we're already
testing.
Fixes gh-1938
Closes gh-2166
IE10+ may return not exactly the offset.top value set in an offset callback
if parent has fractional top offset itself. Checking for being close to the
desired result fixes the test error.
Fixes gh-2147
- removes descriptor allocation
- restore simplified cache creation
- adds early return from remove call where no data exists
- use Object.defineProperty
- remove unnecessary code path
Closes gh-2119
- avoid explicit data.discard() cleanup calls
- explicitly remove the data.events property, only when private data exists
- reduces code footprint
Fixes gh-1734
Close gh-1428
QUnit update broke couple traversing and manipulation tests, since new qunit
added another form to the test-suite while some of
jQuery selectors weren't specific
Ref 2d5c5d213f
* Remove "async = true" from script transport since it was needed
for FF < 4 and old Opera which we do not support anymore
* Add comment to "evalUrl" method on why "type" field should be explicit
Android 2.3 resize the iframe by its content meaning it's not possible
to scroll the iframe only its parent element.
It seems (not confirmed) in android 4.0 it's not possible to scroll
iframes from the code
Fixes gh-1981
Ref 4ab8603669
This hack turns out to be needed by Android 4.0-4.3.
Add a support test so that the hack is invoked only where needed.
Refs gh-1815
Refs gh-1820
Closes gh-1842
