> Does anyone have a good overview of how well substitution groups and
> redefines are supported in various processors. The last big project where I
> used redefines extensively about half a year ago I had to redo halfway
> through after running into too many problems, problems where the redefine
> was proper and was supported by some processors but failed in others, even
> more insidious where cases where I had redefined incorrectly and it
> functioned in some processors or in some test instances only to fail later.
> This has put me off redefines, now I'm on something where redefines and
> substitution groups are being proposed as the extensibility mechanism. I've
> had misgivings about substitution groups, finding them somewhat overly
> complicated and have thus avoided them. How is their support?

Bryan,

I recently ran through a series of tests on <redefine> in terms of 
validation and was very surprised at its portability. I too had played 
with it a year or two ago and found that support was lacking. I am only 
just getting into use in other tools like data binding and intellisense 
type applications-- so I haven't really tested that.

The only real interop problem I ran into was with <redefine>ing 
chameleon components into a schema with a target namespace. The behavior 
seemed consistent for simple examples. But when I tried to 
simultaneously redefine a chameleon component as it was being included 
in a schema with a targetNamespace and add attributes from another 
imported namespace the validator choked on the instances.

The results were good enough that I was able to recommend its use. I 
also did not test any forms of restriction-- which only compounds the issue.

Cheers,
Jeff Rafter

From mike@s... Fri Jul 08 23:19:16 2005
Received: from mag