Re: [topicmaps-comment] mergeMap Pointing to MoreThanOnetopicMapElement

["W. Eliot Kimber" <eliot@isogen.com>]

Lars Marius Garshol wrote:

> | Which spec (XPointer, XPath, XTM) defines this as being equivalent
> | to:
> |
> | //*[@id='foo']
> |
> | ?
> 
> This is defined by XPointer:
>  <URL: http://www.w3.org/TR/xptr/#bare-names >

Interesting--the referenced section equates bare names to id('foo'). 

This makes sense in that it is the most flexible *if you have DTD- or
schema-aware processing* *and* have a document with a schema or DTD. But
if you don't, then I would think that an XLink processor would be
obligated to fail to resolve a bare name--but I think that the
expectation is that even for DTD-less topic-map documents that a bare
name reference would work (and that's how I've implemented my mergeMap
resolver at the moment).

Of course, if XTM abandoned XLink and defined its own addressing
attribute, it would then be free to both limit the use of XPointer and
define the meaning of bare names unambiguously.

This seems like it might be reasonable change--the syntax hit is
small--really you only need to move the xlink:href attribute into the
XTM name space.

Please note--I'm not being pedantic about the XTM spec in order to
justify any particular behavior of my processor--I'm simply trying to
understand what the exact correct behavior should be and there seems to
be either ambiguity in some of this and/or disconnects between what the
specs say and what people will likely expect in some areas.

Cheers,

Eliot
-- 
W. Eliot Kimber, eliot@isogen.com
Consultant, ISOGEN International

1016 La Posada Dr., Suite 240
Austin, TX  78752 Phone: 512.656.4139