Re: SV: XSD feature check-lists

Bryan Rasmussen

> Noah Mendelsohn wrote:
> > XML Schema 
> >is no more complex to implement well than languages like, say, Java. 

> This may well be the most damning indictment
> of XML Schema that I have ever
> read. 

...and presumably of XSL as well.  We've built both Schema and XSL 
implementations in my group.  I'd say that getting XSL to run correctly 
might be just a bit easier than getting Schema to run, but the difference 
isn't huge.  Conversely, I'd say that getting an XSL system to run a wide 
range of cases interestingly fast is significantly more difficult than to 
get a schema implementation to run interestingly fast.  I've never built 
an actual Java runtime, but my guess is that XML Schema is actually much 
simpler, except perhaps if youi're doing a really toy interpreter for 
Java. 

I'm not trying to justify unduly complex technologies.  I've already said 
that I thought the schema language could have been and should have been 
simpler, cleaner, more orthogonal, etc.  I am noting that there are other 
XML technologies of broadly the same complexity as Schema that are well 
received.  Yes, XSL is in some respects better designed, and the 
documentation on it is much more approachable.  Java came up because you 
seemed to imply that systems of the complexity of Schema suggested a need 
for managing subset implementations;  I was using Java as an existence 
proof of an even more complex system in which such subsets are not 
tolerated (either legally or technically) and have not been shown to be 
necessary.  For that matter, I don't think the community would be very 
happy if XSL implementations dropped features that proved to be work to 
get right (priority?)

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------

Received on Tuesday, 9 August 2005 06:28:15 UTC