RE: ISSUE-16: Multiple patterns for a single data structure

> It seems apparent that we will have structures identified in our roadmap 
> which may be exhibited in Schema using more than one pattern. 

> Should we treat all such patterns equally, or should we recommend some 
> patterns as being preferred over the rest?

I see this as being tricky, hence raising the issue :-)

Ed's Null type is a good example. How do we evaluate ASN.1's
pattern as being somehow *better* than XMPP's?
http://lists.w3.org/Archives/Public/public-xsd-databinding/2006Jan/0050.html

The same applies to the alternative open enumerated type, 
offered by Pete, over the union based version:
http://lists.w3.org/Archives/Public/public-xsd-databinding/2006Jan/0049.html

> Should we even cite one pattern as being the canonical pattern for a 
> given data structure? 

I think this has value, so long as we don't fall
foul of discouraging the other, valid patterns, *but*
this does expose the relationship between a pattern likely
to lead to a good user experience with databinding tools, 
and a pattern having to be mapped onto a particular structure
regardless of the processing environment.

Paul

Received on Monday, 30 January 2006 14:32:53 UTC