Re: [xml-dev] To do list relative to W3C XML Schema

[Eric van der Vlist <vdv@dyomedea.com>]

Hi Ronald,

Ronald Bourret wrote:

> Eric van der Vlist wrote:
> 
>>Just wanted to recap here some ideas and questions related to W3C XML
>>Schema which, I think, would be interesting to pursue.
>>
>>1) Little intrusive XML PSVI
>>http://lists.xml.org/archives/xml-dev/200110/threads.html#00556
>>
> 
> This is a nice idea. 


Thanks!

> Unfortunately, I see a couple of problems with it:
> 
> 1) The result of a schema-validated document is no longer valid. This is
> because the PSVI information wasn't specified in the original schema and
> it is not realistic to expect people to include PSVI items in their
> schema.


I am not so much expecting people to include PSVI items in their schema, 
but rather validators to do so. And having a PSVI which would be as much 
"schema agnostic" as possible would allow to get some kind of 
compatibility between the output of different schema languages (even 
those which could be "home made" and implemented as a SAX filter or a 
XSLT transformations designed for a specific application and able to 
check all the "business rules").

This would have the benefit of making PSVI "real": I think that it would 
be much more concrete for people if they could "see" it within angle 
brackets!

In this context, it would be more "elegant" if the document with PSVI 
was still valid, but I don't see it as a major issue. If people want to 
validate the document again, the stripping of the PSVI information could 
be done by a trivial XSLT transformation or SAX filter.

 
> 2) This requires all tools that accept schema-validated documents to be
> smart enough to ignore the PSVI elements/attributes. It's not clear that
> this is that much better than having to rewrite those programs to treat
> (for example) xmlns attributes as namespace declarations, not
> attributes.


That's a real issue (and that's why I haven't qualified this proposal as 
"non intrusive").

It will not be transparent, but I don't see how we could make it 
transparent with XML 1.0, namespaces and XPath being what they are.

It could be non intrusive if the PSVI info was added using PIs (which 
might not have been a bad idea since a PSVI is litteraly a Processing 
Instruction) but you wouldn't be able to get the info easily into XPath 
expressions.

It would be almost non intrusive if the PSVI info was added using 
attributes, but would wouldn't be able to add "attributes to attributes" 
and I don't think it can fit either.

My bet is that we can still live with a "little intrusive" PSVI and give 
some guidelines on how to write applications for which this is transparent.

Thanks

Eric


> 
> -- Ron


-- 
Rendez-vous à Paris pour le Forum XML.
                    http://www.technoforum.fr/Pages/forumXML01/index.html
------------------------------------------------------------------------
Eric van der Vlist       http://xmlfr.org            http://dyomedea.com
http://xsltunit.org      http://4xt.org           http://examplotron.org
------------------------------------------------------------------------