Re: Comment on XPath

On Tue, 12 Oct 1999, James Clark wrote:

> Rick Jelliffe wrote:
> 
> >     1) there is no ID axis
> 
> I'm not sure what exactly you have in mind for an ID axis.  I'm guessing
> it's an axis that would be non-empty only for an attribute node that was
> declared as IDREF or IDREFs, and would contain the element nodes
> referenced by the attribute. Is that right?

Yes. Perhaps it would be better to call it IDREF axis. 

Sorry this comment was so late; probably too late for this version of 
XPath, but I hope not.

I gave a paper last week "On Validating Webs rather than Trees" and 
consequently looked through several of the WDs at W3C, and it seems a 
systematic weakness of W3C specs that everything is trees: it seems that 
graphs are regarded as something that happens in the future when XLink or 
XML Schema or XML References whatever arrives. DOM, XPath, XSL, even the 
XML Schemas seem to be based on trees.

This was quite a surprise for me: it seems such a step backwards from 
even 1986 SGML. I had thought that the RDF-ish move towards less deep 
element structures with more use of linking might make treeloc less 
useful and graph-navigation more important. Because of ID/IDREF, SGML was 
never trees but always (untyped) graphs: when we were all discussing 
"SGML on the Web" and the idea of having no DTD came up, it never occured 
to me that the tree model would become regarded as the basic case, and 
that the local graph would end up being put in the too-hard basket in the 
specs that build on XML.   


Rick

Received on Tuesday, 12 October 1999 13:21:21 UTC