> Andrew Ho wrote:
> >If you show us some examples of differences in XML schema, I think 
> >your work would be easier to understand and implement.
> Thomas Beale wrote:
> the XML-schemas will start emerging in the next week or so. We have
>  done the data types. Still have to do the others.
Still as noted in your recent docs XML Schema might not be particularly
suited to the job. 

Quote from openEHR EHR Ref Model rev 2.81, p16
  XML-schema is not well adapted to information modelling ... because
  "it is purely data oriented and missing a number of important
semantics
  namely constraints, generic types and multiple inheritance."

I've come to that same conclusion studying its alternatives:
RelaxNG and Schematron which may provide more practical assistance
while being less verbose/declarative.
There's a well reasoned tutorial about this at:
http://www.topologi.com/public/Schtrn_XSD/Embedded_Schematron.html

In any case schema languages are only abstraction of the part of
the loop.
Some where in that loop some software has to know either
1) what to do if it receives badly typed data - i.e. handle exceptions
2) filter data on the client to prevent badly typed data.
Is it going to be feasible to auto-generate / compile such stuff
from Schema?



-- 
Gavin Brelstaff - BioMedical Area, CRS4 in Sardinia

Reply via email to