Lists Home |
Date Index |
At 11:49 AM 9/27/2002 +0100, Norman Gray wrote:
>Would XLink 1.1 be reasonable? HLink and XLink, though they share the
>underlying ideas, are syntactically very different. It doesn't seem
>sensible to ram both of them into the same specification.
I can see your point. However, the concepts work together. XLink
was intended to have an attribute remapping aspect, we just couldn't figure
out how to do it. Remember, these were back in the days of "pure XML" when
it was viewed as the panacea for all problems. From my perspective, the
XLink WG would have gotten slapped down had they suggested an approach like
The political climate in the XML world is a little different these
days. I think the concepts of HLink could be subsumed into XLink.
>Would it not be better, therefore, to drop the current XLink completely
>and simply call HLink `XLink 2.0', since it doesn't really seem to
>have too much to do with XHTML.
No, I doubt it would work. A lot of us XLink fans would probably
be a little upset. It also doesn't make any sense, since HLink's semantics
are based on XLink, not separate from it. Do you see any definition of
'actuate' in the HLink spec? Me, neither.
>and know what it's expected to do with it. However, your description
>of it as merely `really quite nice' suggests that there's a use-case
>here for which XLink's practice of scattering special attributes is
>an adequate solution. What am I missing?
You're missing people like me, who like linkbases, and links that
are embedded right in the document. Attribute remapping has always struck
me as messy. Necessary, but messy. For example, let's say I want to style
those links (i.e. add behaviour): instead of a processor going through the
document, finding the link, recognizing it, and styling it, we now have to
process through an attribute remapping definition, then through the
document. Sounds awfully similar to validating a DTD to me; what if I just
want well-formedness, and to parse the links directly through a link
processor and a style processor?
People seem to have the opinion that XLink was never meant to
include attribute remapping, or that the group just ignored it. Neither is
the case. Attribute remapping was one of the original goals; and we did a
lot of work on it. We just couldn't figure out a way to do it that would go