OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Non-deterministic content model



So now we would have a Schema-Lite?

Sounds like XDR.

The notion of using different schemas at different 
points in production is a good concept and was valid 
even when only DTDs were used.  A lot of the work 
on modularity prior to webSGML arose from the notion 
that one needed base modules from which one could 
assemble DTDs for different stages of production, 
restricting and relaxing rules at each stage.  There 
was also fancy work with OR grouped DTDs to enable 
the editors to take advantage of bottom-up definitions.

As to having different schema languages, that is a 
bit tougher because now one gets a fatter object 
framework and that is a maintenance/cost problem. 
My opinion is that until the designers get more 
practice with using XML Schemas, working out 
modular designs, and adding rule constraints 
such as normatively might be created with 
XSLT sheets and practically with Schematron, 
it is too early to look at alternatives. 

Cost is the driver.

Len 
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h


-----Original Message-----
From: Rick Jelliffe [mailto:ricko@allette.com.au]

This last one is worth keeping in mind, it may be that each stage in an
information flow is optimal for a different schema language (or paradigm!).
So if one has limited resources, one probably picks whichever schema
language or paradigm will have the broadest impact, knowing the solution is
less than perfect in each stage. But if one has a few more resources,
creating stage-dependent schemas might be worthwhile.

For example, one might have one DTD for accepting data from contractors and
data conversion houses, but a much stricter one for new documents
created -house.  Or you might use XML Schemas for autogenerating Java
interfaces, but use Schematron to check that incoming data is consistent
with more complex rules.  Or you might use RELAX for your XSL validation,
but DTDs for the resulting documents.

I don't expect there will many people using XML Schemas -and- RELAX -and
XDR/SOXetc. in the same house.   Personally, I find it very difficult to
keep different grammar conventions in mind.

Perhaps we need a "Common Structures" proposal, which would provide the
union of XML Schemas and RELAX NG, probably using XML Schemas syntax. Like
RELAX Core, I guess. Kawaguchi-san's article on XML.com is probably a good
place to start (without endorsing it.) Any volunteers?