[
Lists Home |
Date Index |
Thread Index
]
[Erik Wilde]
> Paul Prescod wrote:
> > Erik Wilde wrote:
> >>so what about defining an xlink data model in terms of infoset
> >>extensions, referencing it in xhtml and defining an xhtml-happy syntax
> >>for it, and then going ahead and defining a css3 module for link
> >>formatting?
> > So what you're saying is that XHTML and XLink will be hard-coded to
> > integrate with each other at the prose specification level (rather than
> > through any syntax) and other vocabularies will have to integrate with
> > XLink through the intrusive XLink-namespaced attributes? I don't see
> > that as progress.
>
> it's not the prose level, but i don't have any complete solutions yet i
> could pull out of my pocket. i think that the xlink data model would be
> defined in terms of infoset contributions, and that this spec would
> define a namespace uri. this namespace uri would then be used in the
> xhtml spec to refer to the semantics and infoset contributions defined
> in the xlink data model spec.
>
Great heavens, you are not suggesting having a standard data model
***before*** designing the system, are you??? How confining! How unlike
the development of markup!
Seriously, I'm a model-first person myself, but I have come to see
advantages in having markup be syntax-only so far as possible. This puts me
in a bit of a quandry with respect to your suggestion, which I take to be
the ideal of a generalized model rather than specifically "infoset
extensions".
This is the dichotomy I see in this thread. Should there be a unified
underlying linking model which can then be expressed in some kind of syntax,
or should there be a standard syntax for linking but no underlying
semantics? The traditional Markup Way, expressed by several posts already,
would seem to be the latter.
Let's explore this a bit more. What should be the standard underlying
semantics, if there were to be any? Forget the infoset for now, which
perhaps is a hybrid between markup syntax and semantics. What standard
things would be useful for the potentially infinite variety of markup? Are
there any? Why prefer the first or the second approach?
Cheers,
Tom P
|
