• From: "MURATA Makoto (FAMILY Given)" <eb2m-mrt@a...>
• To: xml-dev@l...
• Date: Tue, 06 May 2008 14:51:33 +0900

Noah,

> I don't think you've accurately characterised the XSD language.  

Roger quoted what I wrote.

> You can, 
> to a significant degree, to incremenal validation if your validator 
> supports it. 

Here we apparently disagree.  I think that W3C XML Schema fails to 
meet the goal.

>  First of all, you can get a degree of modularity by using 
> facilities like xs:include.

No, xs:include and xs:import merely provides modularization of schemas. 
Modularization is a good thing, but you still have to understand a lot 
about what you want to include or import

I would argue that include/import is simlar to procedure calls 
while NVDL and its precedessors are similar to software components.  

In the case of NVDL, authors of component schemas have to understand 
nothing about the overall picture.  In WXS and RELAX NG, authors have to
understand everything.

You might be interested in my recent note (see my another mail 
"Full validation of Atom feeds containing extensions").  I believe 
that W3C XML Schema cannot capture advanced examples shown in this note.  
RELAX NG probably can do that, but I do not want to maintain such 
complicated schemas.  NVDL and its predecessor NRL provide practical 
solutions.

Cheers,

-- 
MURATA Makoto (FAMILY Given) <EB2M-MRT@a...>