[sc34wg3] XLink support in XTM

[Lars Marius Garshol]

* Conal Tuohy
>
> I'm not sure if know what Murray intends to include in that term,
> either, but if it includes, for instance, this section (about the
> xlink:href attribute) from the XLink spec, then I'm all for it:
>
> http://www.w3.org/TR/xlink11/#link-locators
>
> It references IRIs, how to escape special characters in URIs,  
> xml:base,
> and XPointer.

To get these things for free by referencing XLink is an argument that  
makes sense. (For the record: I have no idea if this is what Murray  
meant or not.)

> All of this, I think, is worth having.

Well, let's go through these one by one:

   - We already reference IRIs, and can't escape that whether we use  
XLink
     or not. So no win there.

   - The escaping of characters in URIs we may need, and may decide  
to solve
     by referencing XLink (as opposed to actually using the XLink  
namespace).
     I'll double-check this.

   - We don't use xml:base, so no win there.

   - We decided long ago not to use XPointer for references  
internally in
     XTM, so no win there, either.

> I'm absolutely convinced that XTM 2 should conform to this  
> specification.

Are you still convinced about this after my reply? If so, why?

> I would certainly be appalled if XTM 2, for instance, handled xml:base
> differently to the XLink spec, or simply failed to be specific  
> about it.

I would agree. That would be an unacceptable bug.

> IMHO there's no better way to indicate this conformance than to  
> include
> the XLink spec by reference.

Well, merely including XLink by reference is not enough. We need to  
also use the XLink namespace, make clear where we require extra  
behaviour/semantics compared to what XLink does, etc etc

> I realise that much of XLink (extended links) is irrelevant to XTM,  
> but
> I think so long as the XTM spec makes it clear that XTM links are ALL
> "simple" XLinks, then all of that stuff (extended links, locators,  
> arcs
> and roles...) needn't be seen as a complication in the XTM spec  
> itself.

It's not that simple, I'm afraid. I felt I explained this quite well in

   http://www.isotopicmaps.org/pipermail/sc34wg3/2006-March/003151.html

Note the first and third bullet points especially. This is pure  
overhead that XLink adds, and I've yet to see a single thing that  
XLink offers us in return that is worth the overhead of using the  
XLink namespace. (Referencing XLink to say that you must escape URIs  
the same way is fine; there's no overhead in that.)

--
Lars Marius Garshol, Ontopian               http://www.ontopia.net
+47 98 21 55 50                             http://www.garshol.priv.no