dijit.Tooltip does not show up in Safari with 1.2.0rc2

[jayantbsai]

http://dojotoolkit.org/forum/dijit-dijit-0-9/dijit-support/1-2-0rc2-dijit-tooltip-not-working-safari

[bill]

Hmm, looks like the onMouseEnter/onMouseLeave changes from [14530] (or onMouseEnter/onMouseLeave events themselves) aren't working.

[Tom Trenka]

Looks like Webkit does not support onMouseEnter/onMouseLeave. Probably we're going to have to revert that to onMouseOver/onMouseOut with event bubble cancellation. Any problems with that? This bug is probably the last bug to be truly addressed before cutting the 1.2 release.

[Tom Trenka]

Looking into it, I see that event is being synthesized, we'll figure it out.

[James Burke]

peller seems to have isolated this to this changeset: ​http://bugs.dojotoolkit.org/changeset/15273/dojo/trunk/_base/event.js

On further investigation of this change, it seems very hazardous to route Safari through all the event normalization and special cleanup for the dojo.keys fixes.

[bill]

Yah, I take back my above comment. As peller said it's due to [15273] which fixed some keyboard handling for safari but apparently broke onMouseEnter/onMouseLeave (which was otherwise working fine).

[bill]

Here's the problem: IE is the only browser that gives us native onMouseEnter/onMouseLeave support. We fake it for the other browsers (see the definition of add() at the beginning of event.js):

add: function(/*DOMNode*/node, /*String*/name, /*Function*/fp){
	...
	if(!dojo.isIE && (name == "mouseenter" || name == "mouseleave")){

But now safari (along with IE) is overriding that definition of add():

// IE event normalization
if(dojo.isIE || dojo.isSafari){ 
		...
		add: function(/*Object*/ source, /*String*/ method, /*Function*/ listener){

thus losing the code to fake onMouseEnter/onMouseLeave.

[bill]

(In [15348]) Fixes #7748, refs #6876: reinstate faux onmouseenter / onmouseleave events for safari. !strict