Re: The Semantic Debate

On 5/7/07, Jonas Sicking <jonas@sicking.cc> wrote:

> I would not be opposed to adding a 'role' attribute, as long as we also
> support adding semantics the way it's done before. Such as using the
> class attribute (as long as it's properly prefixed as has been suggested
> before) or using new elements added to the spec.

(1)
Since class is officially meaningless, we it is impossible to break
any officially sanctioned prior usage.  People who were careful enough
to consistently use a private meaning in the past will probably be
careful enough to update their documents (if required) before updating
the doctype.

(2)
That said, it would not be unreasonable to use role (or some new
attribute) instead of class, simply because it hasn't already been
defined as officially meaningless.

(3)
Using class=(previously invalid value) has the worst of both approaches.

Because it is an existing attribute, there is a chance of clashes.
(If only with tools that arrogantly assumed they would have full
control of the class attribute.)

On the other hand, since the value isn't valid today, it isn't the one
people are already using.  Even if the authors themselves say that the
new meaning is correct, they still have to edit their document and
change the value to something more obscure before they can benefit.

-jJ

Received on Monday, 7 May 2007 21:16:38 UTC