> >Well-defined semantics for at least some values of
> > REL would begin to make it useful, and then it just might be taken
> > seriously by document authors.
>
> I disagree. Standard semantics only get in the way. The strength of
> the REL mechanism is the ability to build a completely custom set of
> link types that match the structure of the document... I agree that
Seems to me I'me seeing two orthoginal applications of the REV/REL
attributes. I'm seeing the suggestion that they be used by browsers to
impose a navigable structure on a collection of pages and I believe that
there must be a well known set of terms that most well behaved browsers
will implement in a fashion that content providers can depend on or they
won't bother.
On the other hand, I believe Craig envisions that there is no predefined
UI behavior semantic expected. Rather the browsers provide users with
convenient access to the values for human interpretation of semantics.
In that context I would agree that predefined terms are not required.
My conclusion would be that both are valid objectives. But either we need
different attributes or perhaps we could allow some form of attribute value
qualification like:
rel="comment(Dave's book) parent"
where "comment" is a well known term meaning display to the user this
interpretation of the relationship. While "parent" is a well know term which
browsers would use to add semantics to their management of document history.
Dave Morris