timmywil [Fri, 25 Mar 2011 04:35:50 +0000 (00:35 -0400)]
Fix #7472 and added test for #3113
- Forms with an input that has either name="action" or name="some-other-attr-on-the-form" caused problems in IE6/7. This is fixed.
- Changed check in $.attr for ret === null to typeof ret === "object" to catch any inputs that are accidentally retrieved in IE6/7, since attributes cannot be set to objects and typeof null === "object"
timmywil [Sun, 13 Mar 2011 19:44:51 +0000 (15:44 -0400)]
Full test suite now passes in all browsers! There are probably some tweaks we can make to shorten and simplify.
- removeAttr now only uses setAttribute if camelCase setAttribute is not supported
+ Might want to rename jQuery.support.getSetAttribute
- tabIndex is a special case now for hooks where undefined should be returned.
+ Should we be checking if hooks returns undefined? undefined might be the desired return value in future hooks.
As of now, tabIndex is the only one that needs it, but the test suite will still pass if we don't check if hooks are undefined.
jaubourg [Wed, 30 Mar 2011 16:54:32 +0000 (18:54 +0200)]
Fixes #8692. Strange FF4 bug: values changed onto the arguments object sometimes end up as undefined values outside the $.when method. Cloning the object into a fresh array solves the issue.
Carl Fürstenberg [Thu, 17 Mar 2011 18:50:53 +0000 (19:50 +0100)]
Merge when updating submodules on make
When running make, the submodule update will remove all local changes.
Adding flag --rebase or --merge does solve the issue. rebase will
probably make it cleaner, but it might stop on conflict, thus --merge
will result in fewer (probably none).
Carl Fürstenberg [Tue, 15 Mar 2011 19:16:09 +0000 (20:16 +0100)]
Introduce submodules, closes #8536
Instead of the manual clone of sizzle and qunit, use git submodules
instead. this will ensure that all future releases can be recreated by
checking out an tag.
jaubourg [Tue, 15 Mar 2011 18:20:03 +0000 (19:20 +0100)]
Fixes #8509. Makes URL regexp less overzealous and ensures it recognizes URL schemes which do not contain a conformant hierarchical structure ( as per section 2.1.2 of http://www.ietf.org/rfc/rfc2718.txt ). Also adds about: and adobe air's app: and app-storage: to the list of local protocols and provides a failover in case document.location is illformed. Unit test added.
Dave Methvin [Sat, 5 Mar 2011 02:16:40 +0000 (21:16 -0500)]
Fixes #7340. Use a single capturing handler to simulate bubbling focusin/focusout event on non-IE browsers. Allow native DOM methods to fire events other than the currently active one back into jQuery.
jaubourg [Thu, 3 Mar 2011 18:38:06 +0000 (19:38 +0100)]
Moves Deferred-related code into a separate module. Context handling has been simplified in _Deferred.resolve and jQuery.when has been refactored for clarity (and minor optimization).
jeresig [Wed, 23 Feb 2011 18:18:44 +0000 (13:18 -0500)]
Make a new jQuery.support.noCloneChecked - splitting apart the previous feature detect relating to clone in IE, fixes the last remaining issue with IE 9 RC. Fixes #8365.
jaubourg [Fri, 18 Feb 2011 17:06:26 +0000 (18:06 +0100)]
Revert "Adds an invert method to promises that returns a "inverted" promise that is resolved when the underlying deferred is rejected and rejected when the underlying deferred is resolved."
Anton M [Thu, 17 Feb 2011 18:12:57 +0000 (19:12 +0100)]
Fix a some inaccuracies in the original test case for #7912.
- Use fresh div instead of one outside the test-fixture
- make sure the empty string test tests actually that (not 0% 0%)
- actually test for < -10000 (#7193)
- fixed some whitespace issues
louisremi [Thu, 17 Feb 2011 16:26:23 +0000 (17:26 +0100)]
Fixes #7912. Make sure .cur() only returns 0 as fallback value when it needs to ("", auto, undefined, null).
This change makes .cur() more .cssHooks friendly. .cur() now returns the unmodified value by
.css() if it isn't a number, number-alike or a value that needs a fallback to 0.
This way fx.start doesn't need to be recalculated for complex values.
jaubourg [Thu, 17 Feb 2011 16:03:09 +0000 (17:03 +0100)]
Fixes #8297. Makes sure response headers with empty values are handled properly and do not prevent proper parsing of the entire response headers string. Unit test amended.