This is prettly what I suggested in my last letter -- split the problem
in two: REL/REV to define the relationships, and METHODS (I called it
ACT, but what the hey) to define the operations that can/might be
implemented by the browser.
> given the above, one can already point to a style sheet for links with
> <LINK HREF="..." REL="linkstylesheet"> where "linkstylesheet"
> would need to be standardized. But that's the rub. And Craig's
> proposal rests on standardizing a syntax within attribute values,
> one which can't be checked by SGML parsing.
>
> That's always a red flag to me. It tells me that I'm trying to
> do too much within SGML.
>
> In the case of HTML, it would be easier simply to define a new element,
> LINKSTYLESHEET or something, with the attributes HREF and URN. The
> semantics of LINKSTYLESHEET are that it points to a specification
> for how to interpret the values of REL in LINK. You may want to
> have more than one LINKSTYLESHEET in your HEAD, so make it
> LINKSTYLESHEET*.
I agree completely, and said something similar in a previous letter.
Stylesheets are very different from LINKed resources. Stylesheets
define a document's presentation, and are really a 'part' of the document\
itself. LINK, on the other hand, can define things like related
indexes, glossaries, documents, and so on, which has nothing to do
with stylesheets. The two concepts mix badly, which might explain
some of the problems we've been having.
> But I think that linking to exterior resources such as style sheets
> is better done in something like an SGML catalog. After all,
> the browser should know how to interpret the document's markup
> before it encounters anything beyond the doctype decl. You can do
> the same thing with processing instructions (Panorama does this),
> but PIs present the same standardization problems.
Well..... this may be nice, but HTML is supposed to be simple, and
would this not make it rather more complicated (particularly to 'vi'
hackers like me ;-) ) . I still think that a LINKSTYLESHEET
element, in the HEAD, would work well, be easy to understand,
and easy to implement.
> In HTML, it is appropriate to agree on common values for the
> core functions already implemented. Let's do that for REL, and
> if we want to point to something that explains link semantics,
> define a new element.
Yeah, verily!.....