Jörn Zaefferer [Mon, 21 Dec 2009 21:10:21 +0000 (16:10 -0500)]
Stop trying to emulate the focus/blur event in IE, doesn't work as one might expect, anyway. Instead, implement the focusin/focusout events in all other browsers - which creates a much better parity across all browsers. Uses event capturing instead of bubbling to make it happen. Thanks to Alexander for the recommendation and to Joern Zaefferer for the original focus/blur delegation code.
Alexander Farkas [Mon, 21 Dec 2009 20:32:32 +0000 (15:32 -0500)]
Used the patch from Alexander as the basis for a rewrite of the IE change event logic. Now has full parity with the regular change event in other browsers: Works with regular bind, works better with multiple selects, works as a regular change event (note test suite changes), works with readonly/disabled inputs, and much more. The original patch had a number of problems, including firing the change event too many times, not bubblinb properly, and not handling clicks on multi-selects properly - that should all be fixed now. Thanks Alexander for the patch pushing in the right direction.
jeresig [Sat, 19 Dec 2009 02:23:26 +0000 (21:23 -0500)]
Switched from using YUI Compressor to Google Compiler. Minified and Gzipped filesize reduced to 22,839 bytes from 26,169 bytes (13% decrease in filesize). Sizzle copyright was merged into the main header (since it's removed automatically). Still passes all unit tests.
jeresig [Fri, 18 Dec 2009 17:41:53 +0000 (12:41 -0500)]
Disabled the passthrough .attr(method_name) functionality. You can now use it if you do: .attr({method_name: value}, true) OR as an easy initialization method: jQuery('<div/>', {html: '...', id: 'test'}).
jeresig [Fri, 18 Dec 2009 16:30:04 +0000 (11:30 -0500)]
Backing out the change from 841f9ff7a1815b521044aeeb39ccbe70fa688201, it was ineffectual in Internet Explorer. Keeping the broken test case, for now, so that we can check into the issue some more.
Dave Methvin [Thu, 17 Dec 2009 02:02:58 +0000 (10:02 +0800)]
Add new html5 input types to list of serializable types. Older browers handle these as type="text" so they should be consistently serialized on both old and new browsers. Fixes #5667.
Ben Alman [Wed, 16 Dec 2009 12:41:56 +0000 (20:41 +0800)]
Reorganized param code slightly to once again define buildParams as a named function expression, as moving it outside the param closure would necessitate moving and rewriting the add function as well.
Ben Alman [Wed, 16 Dec 2009 03:54:12 +0000 (11:54 +0800)]
Further modified jQuery 1.4 .param() method to serialize arrays containing non-scalar values. PHP and jQuery BBQ can both deserialize these structures, while rack (as of 1.0.0) cannot.
jeresig [Thu, 17 Dec 2009 19:17:05 +0000 (14:17 -0500)]
Revert the changes from 65ebf57c1e5d7fa96536b66d4fcacbafad8dc1e5, they were ineffectual against IE's insanity: http://ejohn.org/files/bugs/fragment/ Used a different tactic instead (just don't return a fragment from .parent()). Fixes #5638.
John Resig [Thu, 10 Dec 2009 17:25:25 +0000 (09:25 -0800)]
Moved some methods around inbetween core.js and traversing.js. Core methods shouldn't rely upon other modules (e.g. the selector engine) wherever possible.
John Resig [Thu, 10 Dec 2009 04:58:49 +0000 (20:58 -0800)]
Make sure that the defaultView exists before attempting to get at the computedStyle (makes it easier to hide items that don't have a view). Fixes #5619.
John Resig [Wed, 9 Dec 2009 22:49:31 +0000 (14:49 -0800)]
Make sure that we don't try to trigger non-existant native events on applets, embed, objects, etc. as it'll cause an exception with Java applets. Fixes #2414.
Dave Methvin [Wed, 9 Dec 2009 03:46:07 +0000 (11:46 +0800)]
Get browser version rather than rendering engine version.
Don't recognize Chrome as Safari.
See results of different userAgent strings here:
http://spreadsheets.google.com/ccc?key=0Aj5JJFjq9rZDdHdxXzRLVnY0SzFpRTBOUDc4VmhzRVE