Michael Kay wrote:

>>Who needs to dynamically generate schemas? The whole point of 
>>schemas is 
>>to be a widespread, well understood description of instances.
>>    
>>
>
>There was someone with that need writing to the list a few days ago. It
>  
>
I am aware of that request for help, but I disagree. You were suggesting 
to her that multiple schemas might do the trick.

>seems entirely legitimate to me to apply different schemas to the same
>document at different stages of a workflow, or for senders of documents to
>apply stronger validation criteria than recipients of the same documents.
>
>  
>
Granted, but does that require a "dynamically generated" schema?

By dynamically generated, I understand this: A clever way of doing many 
matrix multiplications is to use dynamic programming (ordering the 
multiplications once you know all the matrices' dimensions).

A web application may dynamically generate a response for a request.

In the use case you seem to suggest, senders seem to refine schemas to 
achieve tighter checking. If one knows in advance that one wants such 
refinement, one can write the schemas by hand. If not, then finding a 
mechanical transformation for the old schema will not help much.

I guess, the problem for that use case is modularity.

Coming back to "dynamically generated", I think the very fact that 
somebody can write a mechanical transform to generate one schema from 
another hints at enough anticipation of requirements that the original 
schema could have been written in an extensible way in the first place.

regards,
Burak Emir

http://lamp.epfl.ch/~buraq