xml-dev - RE: Irony heaped on irony

[Derek Denny-Brown <derekdb@microsoft.com>]

MSXML tried to take a post-namespace view of all of XML, including DTDs.
This had 2 side effects:
1) prefixed had to be declared (using default attribute values) on every
element where they are used
2) prefix mappings could not differ between the document and the DTD. (and
thus were required to be #FIXED)

Our internal DTD representation did not preserve the prefixes used in the
DTD at all.  For this to work we had to unambiguously be able to determine
the meaning of a given prefix.  We scoped all prefix mappings to a single
element decl (and it's attributes).  This is the most conservative approach.
All other interpretations lead to ambiguous cases.  This implimentation
choice meant that the mapping had to be repeated all over the place, and
proved to be very confusing to people.

Similarly, treating a DTD in post-namespace manor, implies that prefixes
should map to a single namespace.  By specifying a content model of (title,
desc) should imply the exact same specificity as (a:title, a:desc).  For
this to be true, there must be one, and only one interpretation of the
prefix 'a', thus our requirement that the appropriate xmlns:a default
attribute (since there is no other way than a default attribute to specify
the mapping), must be fixed.


For the latest msxml web release, this logic has been completely reworked.
We now validate against a DTD based on the prefix/localname pair, rather
than the namespace-uri/localname pair.  This puts msxml in line with the
other xml parsers.

-derek

> -----Original Message-----
> From: Julian Reschke [mailto:reschke@muenster.de]
> Sent: Thursday, May 18, 2000 1:15 PM
> To: Andrew Layman; ht@cogsci.ed.ac.uk; Tim Bray
> Cc: xml-dev@xml.org
> Subject: RE: Irony heaped on irony
> 
> 
> Could you please explain what you mean by "very conservative 
> interpretation
> of the namespaces spec", and what you plan to change???
> 
> jr
> 
> > -----Original Message-----
> > From: owner-xml-dev@xml.org 
> [mailto:owner-xml-dev@xml.org]On Behalf Of
> > Andrew Layman
> > Sent: Thursday, May 18, 2000 9:04 PM
> > To: 'ht@cogsci.ed.ac.uk'; Tim Bray
> > Cc: xml-dev@xml.org; connolly@w3.org
> > Subject: RE: Irony heaped on irony
> >
> >
> > You could also declare the attribute value as fixed, that, 
> too, would make
> > IE5 happy.
> >
> > (By the way, as mentioned in some earlier threads, the IE5 MSXML
> > very-conservative interpretation of the namespaces spec, which
> > interpretation is the cause of this trouble, will be changed in
> > the next rev
> > of the parser.)
> 
> 
> **************************************************************
> *************
> This is xml-dev, the mailing list for XML developers.
> To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
> List archives are available at http://xml.org/archives/xml-dev/
> **************************************************************
> *************
> 

***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************