[patch][cla] Add NodeList-data to core

[dante]

This is probably the most often requested bit of functionality for users when they are migrating from jQuery, and we don't have a simple solution. dojo.data is "too much", and telling folks to write their own plugin for it is silly. I wrote one, it lives in plugd. It differs slightly from $.data in that you can pass a hash as a setter, and get an entire data object back by passing no parameters.

Exposes a single-node function dojo._data (which probably needs renamed, but is private ... )

One thing to note, and I'm note sure if I like it: Instead of following the other NodeList functions that return a map, I found the single-node case to be the most relevant. eg: dojo.query("#id").data("bar") ... most other NodeList functions return an array from .map, but I made this one determine if it is a single item list and return !this[0]. Feels like magic, but the alternative is cumbersome:

// setter:
dojo.query("#bar").data("bar", { a:"b" });
// getter, traditional:
if("b" == dojo.query("#bar").data("bar")[0].a){}
// getter, magical:
if("b" == dojo.query("#bar").data("bar").a){}

The magical version being not so magical and failing violently when your list is .length > 1 and you aren't expecting it to be.

Thoughts? The inline docs reflect the non-magical state, and will be updated to reflect whatever is decided.

[dante]

worth pointing out, lots of the tests pass in DOH because it seems arrays are cast when tested with doh.is? eg:

dojo.query("#fooli").data("bar", "baz"); 
var info = dojo.query("#fooli").data("bar");
// in non-magical case, return is ['baz'], magical is: 'baz'
doh.is("baz", info); // passes in magical case because ['baz'].toString() == 'baz'

threw me for a loop when going back and forth on if it should be magical or not.

[James Burke]

Looks good to me. From the meeting it sounds like it would not have the magical behavior of if only one node, return data object, but always return an array. I would like that to change in the future though, perhaps in a 2.0 thing, but understand keeping consistency with the other NodeList? get operations.

[dante]

(In [22788]) fixes #11637 - adding NodeList?-data to core. docs are ready, no live examples until after commit ​http://docs.dojocampus.org/dojo/NodeList-data

[bill]

In [29416]:

don't call doh.t() with a string, it tries to eval it, making a spurious test failure about a syntax error, refs #11637