Well, that depends. I would say [...] _encapsulates_ an attribute
specification. The end of the attribute value might otherwise need to be
terminated by a ) or a , (context sensitivity or grouping, respectively).
>> Ah, but the class attribute specification is obviously not meant to
scale.
>
>Why not?
I didn't mean it *shouldn't* be intended to scale, just that the
specification was obviously a one-shot deal, meant for specifying class and
nothing else.
>> ? I thought there was already a "proposal" for generalized attribute
>> selection - the [attribute=value], or just [attrib] for existence,
format.
>
>Yes. I was saying that some of the tokens in there are superfluous. I made
>another proposal. This also had the advantage of being clear that a.b was
>a short form of the generic syntax I proposed, which can be applied to
other
>SGML DTDSs in addition to HTML. It regularises things.
Ah. Merely a different syntactic proposal.
>> That was, I believe, the point of making "[ID=value]" so complex in
>> comparison (I don't know why we didn't just do "[CLASS=foo]" from the
>> beginning - perhaps we should?
>
>To make them more consistent? Fine. Although, as I say, the trailing
>square bracket is not doing anything.
I'd still vote for it, the same way I wish <LI> were forced to be a
container.
-Chris