.attr() proposal

[flesler]

I really think something is not right, with the way attr works. You need a huge hash ($.props) to store names, where most of them are xyz: xyz, still, many users complain that attr doesn't return the expected value.

I'll attach a proposal that gives priority to DOM 0 methods, that are, in the end, oftenly used (scrollLeft, parentNode, etc). It also includes a fix for ticket 2537, where attr returns undefined when it should return 0.

Also slightly optimizes the amount of code.

[flesler]

This is not a bullet-proof patch, more like a draft of what I think it should look like. It surely skips some situations, that were already patched, that needs to be merged in, if this gets included.

With this patch, new things will work like:

$('#foo').attr('parentNode');
$('#foo').attr('selectedIndex',5);

Also, I could now safely remove from jQuery.props, all the foo:foo options.

[flesler]

Related Tickets:  2537,  2521,  2453,  2170,  2119,  2099.

[flesler]

Proposal

[scott.gonzal]

One thing that stood out to me was this line:

get = value != undefined,//whether we are getting (or setting)

which should be:

get = value !== undefined,//whether we are getting (or setting)

in order to prevent mistaking null for undefined.

[flesler]

Thanks for checking, but I disagree. $().attr('foo',null) is not a normal call, null shouldn't be handled like data, but what you say is logic. I'll let others decide on that.

Good catch.

[flesler]

Yeah, the null should considered setting not getting. Will change that if this gets approved.

[brandon]

Nice cleanup patch. Make sure null is setting the value and this is ready for commit.

[brandon]

One more issue I noticed ... we need to avoid the expando if-branch if the node is from an XML document. It would probably be good to go ahead and store the value of jQuery.isXMLDoc(elem) in a var since it is used multiple times within the attr method.

[flesler]

Implemented at [5574].