[Sorry for the late response.] Laurence, This isn't an urgent issue and I agree we should talk more about it, in the context of Ellen's Information Modeling Architecture paper. A call or the OGSA F2F might be a good venue for this discussion. Your response makes me think that I may have the wrong impression of where GLUE 2 is positioned in that architecture. Andreas Laurence Field wrote:
I think that we have a chicken and egg situation here.
Should the Glue schema model what is required by JSDL or should the JSDL contain what information is available?
Personally I feel that this should be user driven and JSDL should define what the user is looking for. The Glue Schema should reflect this and decide the best way to provide this information. The user needs to understand what are the important attributes about the resource to which they have to match. I don't think it is up to the resource provider to decide what attributes they think are interesting to be published. This should all be use case driven and hence dependent on what the user wishes to do. In any case all this needs to be discussed with both groups involved.
Laurence
-- Andreas Savva Fujitsu Laboratories Ltd