Rebecca S. Guenther wrote:
> I haven't heard any comments about this suggestion below.

Hi again Rebecca

I agree Ray's solution is probably the simplest option that makes the 
PREMIS XSD valid again, and does so in the (inflexible but readable) 
style of the original.

However if the true design intention is that these two date elements are 
in fact of the same type (as they probably should be), the logical scope 
for the type definition is global, either from PREMIS.xsd, or a separate 
definitions file that can be included by any of the component schemas.

Thus far I have only referenced PREMIS.xsd in my XML models; the 
flexibility of referencing individual components separately seems less 
valuable to me, than being able to redefine locally those element types 
(EventType, AgentType...) that the PREMIS report itself earmarks for 
local control. Readability of the /raw/ schema also seems a relatively 
unimportant goal - XSLT can deal with that.

Therefore, I would still urge a rethink of the extensibility of the 
current XML Schema implementation of PREMIS, along the lines Ray and I 
have suggested, and echo Jerome's concerns about the risks of 
arbitrarily locking the current implementation down for a year with no 
hope of remission.



\ Richard M Davis
/ Digital Archives Specialist
\ University of London Computer Centre
/ Tel: +44 (0) 20 7692 1350
\ mailto: [log in to unmask]