Costello, Roger L. wrote:
> Wanted: persons with the ability to translate schemas into
> schema-language-independent prose...
>   
(I know it is not what you are asking but...) Why not express the 
schemas in schema-language-independent prose in the first place, then 
mark that up?  

Obviously this is the notional modus operandi of a Schematron user, but 
more or less the same thing is available in RELAX NG or XSD if you make 
a convention about what comments are actually used for (for example, and 
have markup that distinguishes between description or links to 
requirements and comments on implementation.)  Since XSD and RELAX NG 
software is not really designed to cope with this assumption, the 
documentation features of XSD and RELAX NG have never gone anywhere, 
unfortunately.  The difference is of course that XSD and RELAX NG have 
nice slots to talk about particular components, but no slots to have 
documentation on the relationship between components (if element X has 
to come after element Y, why is that?)

The current form I am encouraging for Schematron assertions is this:

    An X should have a Y because Z.

In other words: context/constraint/trace  where we need to be open to 
the possibility that any context that never exists in fact, or 
constraint that cannot be traced to a requirement, or requirement that 
cannot be tested, may be bogus or a cause for investigation.

Cheers
Rick Jelliffe