On Mon, 30 May 2005, Ernest Cline wrote:

> While the semantics of what exactly a <dl>
> is and should be used for is a matter of some contention, they
> serve a semantic propose that cannot be adequately handled
> by any other simple structure, namely that of indicating sets of
> relationships.

"Sets of relationships" sounds like an attempt at a (vague) semantic
definition. It would be completely different from the definition of <dl>
in current HTML, and in the XHTML 2.0 draft.

> Of course, you'd also have to use a goodly quantity of CSS to
> get anything approaching the default presentation of a <dl>,
> but that is of secondary importance to the semantic of
> relationship pairs.

What is the semantic of relationship pairs? If you defend <dl> on _such_
grounds, I think you should propose an exact formulation of the semantics,
instead of the "definition list" definition followed by prose that tells
it was just a joke. What relations are expressed, and between which
elements? Markup is all about relations; to refer to relationships in
general is semantically empty.

> Whether those pairs are of terms and
> definitions, questions and answers,  roles and dialogs, or
> any other pairs of relationships is of far lesser importance
> semantically.

But that's what _semantics_ is. "Relationship" is abstract, or
metasemantics. We could define the "semantics" of _any_ element by saying
that it describes relationships between its subelements; therefore such a
definition does not define semantics.

The definition should naturally also indicate the semantics of
<dl><dd>foo</dd></dl> and <dl><dt>xyz</dt><dd>abc</dd><dt>foo</dt></dl>
(or say that these do not comply with the specification, on grounds of
some general syntactic rule to be formulated; whether the rule can be
formalized within a particular formalism is less relevant).

-- 
Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/