Michael Kay wrote:
> All I'm saying is, don't make a virtue out of it. The world would be a
> better place if everyone used the same XML validation language, assuming
> that language met all the requirements (and there's no intrinsic reason why
> one language shouldn't meet all the requirements).
I think those are statements of religious belief (or of worldview) 
rather than of fact. Supporting plurality as the basic property of the 
layered web *is* a virtue. The rejection of monolithic unlayered 
standards is how we got our current progress on the WWW.

>  Having multiple languages
> coexisting just adds cost and complexity, just as mixing programming
> languages in an application adds cost and complexity. No-one wants to do it
> if they can avoid it.
>   
Hmmm, I think this is way too simplistic.

Take SQL and Xpath. Are you saying that an application should not use 
both, if that were the most convenient, because doing so would add cost 
and complexity? If the language is small, targeted and high-level 
enough, it reduces costs and complexity. Does JSON add cost and 
complexity (as a necessary fact, merely because it has alternatives)?  

What will add complexity is multiple monolithic heavyweight languages.  
XSD did almost kill off any advances in schema beyond what it could 
support, because it was so heavyweight, but not quite.  Learning or 
using RELAX NG isn't nearly as complicated as learning XSD.  If the idea 
is to reduce cost and complexity, I am not sure that XSD comes 
particularly low on the list of technologies that you would ditch.

Monolithic languages such as XSD make upgrade, evolution, and 
mix-and-match of vocabularies difficult.

Cheers
Rick Jelliffe