Re: General issues, how to replace entities in DTDs using XInclud e.

Jonathan Marsh wrote:

> Thanks for the comments.  I will add them to our issues list for discussion.
> 
> I especially look forward to the discussion of redundancy with entities.  In
> general we have tried to make XInclude compliment entities rather than
> replace them, in the interest of reducing redundancy (under the assumption
> that XML 1.0 entities will never go away).  Your suggestion for full
> rendundancy in order to begin migrating away from entities shows that there
> are two opposing solutions for dealing with the redundancy issue, and now is
> an opportune time to double-check that we're really going down the right
> path.

More comments.

XInclude is usually far from my radar screen.  I am motivated by the 
real world happening around me.  For example, in the XML Protocols 
group, I bring up the nearly-essential role of entities in XHTML and 
other types of XML.  I am told that XML Protocols don't need no stinkin' 
entities because XInclude exists.  The attitude of the SOAP spec towards 
entity references demonstrates that many are trying to make entities go 
away with the advent of XInclude, whether that was a requirement or 
not.  When I bring out the many additional use cases such as character 
entities, parameterization, etc. I get blank stares.  Certain XML 
Protocols members like to think they can embed all XML in the XML 
envelope of XML Protocols, but the simple fact is that frequent reliance 
on entities makes that impossible for common XML such as XHTML.

With a model like the one I described where entity declarations are 
nested inside the elements, it becomes trivial to mingle and nest 
entity-using XML with other XML with no conflicts at all.   I like to 
destroy my own best arguments in cases like these.  This also makes it 
possible to isolate entity processing to namespaced contexts which are 
already of comparable complexity -- the basic XML Protocol envelope 
might still like to avoid entities to make basic processing simpler 
while permitting embedded XML with fancier end processors to encapsulate 
entities and deliver them to the final destination, which if processing 
XHTML, for example, would be entity aware.

It also simplifies the replacement of DTDs with XSchema as I have 
mentioned before.

I appreciate your consideration of the issue.

Best regards,

Ray Whitmer
rayw@netscape.com

Received on Wednesday, 7 March 2001 20:36:23 UTC