RE: [xml-dev] Re: determining ID-ness in XML

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

That's specious.  You are saying that a processor doesn't 
have to support these attributes, but if it is to be used 
generally, it will have to.  Otherwise, system components 
for XML have to be profiled at levels that aren't worth the 
trouble.   

xml:n is a system vocabulary.  It is an orthogonal language 
devised to enable interoperation given different processing 
scenarios, yet it amounts to a set of requirements for 
what has always been a vague notion in xml:  the XML processor. 

In this particular case, we are removing the requirement to 
use standard means of unique identity declaration and providing 
an alternative, yet one which any XML processor will have to 
support to be of general use.   We are conveniently sidestepping 
the fact that some application language designers are ignoring 
standard means with the quaint and altogether illegitimate 
reason that "it sucks".   

ID validation is still validation.  IDREF validation is still 
validation.  Standard means exist and if the application designer 
does not care to use them, they should justify why their choices 
require changes to the definition of the XML processor as realized 
in the extension of the XML system vocabulary.

So far, streaming and XPointer support seem to be the candidates 
for such requirements.  Breaking SOAP is a pity, but the SOAP 
designers chose that and their choices do not put defacto requirements 
on XML or the XML processor.   At some point, the commercial 
users should vote with their feet with regards to the costs 
of maintaining the endless XML experiments.

len


-----Original Message-----
From: Rob Lugt [mailto:roblugt@elcel.com]

Elliotte Rusty Harold wrote:
>
> However, we are not saying that every application has to use xml:id.
Certainly not every application uses xml:space or xml:lang today, much less
xml:base. If the XHTML folks don't want xml:id in their XHTML documents,
they don't have to use it. They don't change their DTDs. In fact, they
really don't need to change them because conformant XHTML documents already
have a DTD, and do have ID type attributes.
>
> Remember, the reason we're having this discussion is because not all XML
documents have DTDs. It is precisely those without DTDs where xml:id is
needed, and these applications can start using xml:id  immediately because
they don't have a DTD to change! or even if they do, they don't require
validity.