overload of class attribute?

[adrienne]

I know this topic has been raised in several places around the web, but i can't find anyone on the mailing list archive speaking to it (this may be a reflection of my searching rather than of anything else).

Is it really a good idea to overload class with structured content? What do you all think of the W3C's very similar proposal to structure content, using the new proposed role attribute?

I think that having a new enumerated attribute seems like a WAY better idea than overloading an attribute that people are already using for a number of other things.

Just hoping to get a discussion started somewhere about this!

Thanks,
Adrienne Leigh Travis

[wgabrie]

You can blame object oriented scripting languages (Javascript, PHP, ... ) for this trend. A class (or role) can be used for style and/or give meaning to the content. The alternative is to add more tags.

[adrienne]

But it doesn't have to be a new tag -- just a new attribute, 'role' or similar. (The W3C's new initiative is using role, that's why i'm using it for my example.

So for instance, my paragraph might look like this:

Code: Select all

<p class="someclass" role="address">Text</p>

See? No overloading.

[wgabrie]

I found a good reason for the introduction of predefined roles in W3's specification. It provides accessibility features. See: Accessible Web 2.0 Applications with WAI-ARIA

Its assistive technology like "Aural stylesheets" but this one is for HTML. Someone should take note of this.

[adrienne]

Right. And it doesn't overload "class" with possibly-changeable semantics. So for BOTH reasons, I think i'd like to specifically propose that this idea get re-examined!

To specifically restate: I propose that WHAT-WG move the "predefined classes" idea into its OWN attribute. "role" is one possible name for the new attribute; I'm sure others can be thought of.

Adrienne L. Travis

[Cerbera]

As I understand it, the ARIA introduction of the XHTML role attribute is only for XML documents at the moment. This attribute could perhaps be added to text/html documents via the work on HTML5, which would be neat. I think this was discussed in brief in the past?

At the moment, IBM's proposal for text/html compatibility does, in fact, overload the class attribute.