Laurent,
> I agree that a CVS or subversion system is very usefull on codes & xml
> documents ...
>
> As I would like us to use the UML model as the reference document, it
> would imply that the reference document will be the xmi document
> which is
> a large & complex document.
>
> In such case, a difference tool would compare instances but I
> imagine that
> the results will be really hard to understand ...
Only if the instances differ greatly, and the version control system has no means to track the history of each file. Modern systems like Subversion and Perforce are built around this need.
> For me, the main problem consist in generating derivated documents
> (web
> documentation, xml schemas ...) from the UML model filled with
> details,
> descriptions, cardinalities ...
>
> Do you agree all to use the UML model as the reference document
> completed
> with all possible details ?
I think that using the data model (XMI instance) as a reference document is a great suggestion. It eliminates the reliance on a single program, and applies to other groups in the IVOA.
However, it would be nice to see the results of an XSLT to create an XML Schema before accepting the XMI version of the data model as a no normative document
--Rick
--Bertrand Russell ------------------------------------------------------------------------ -Received on 2008-02-17Z10:36:32