glue-wg-bounces@ogf.org
[mailto:glue-wg-bounces@ogf.org] On Behalf Of Laurence Field said: I will provisionally suggest that we should have a phone conference on Wednesday 6th of May at 15:00 to discus these initial drafts.
That clashes with a regular meeting for me, which runs from about 14.30 to 16.00 CERN time. I have another consideration to raise, concerning the question of whether the LDAP schema should contain objectclass definitions for the abstract classes. I think if we don't do that it makes life quite a lot harder for anyone trying to prototype an extension to the schema, at least if they want to do it with LDAP. For example, say that you wanted to create a new specialised version of Share to describe FTS channels (which may well be a real use case, currently it's done rather clumsily with GlueService and GlueServiceData). If there is no Share objectclass you have to do all the work of defining and deploying a new FTSShare class in the schema before you can test anything. By contrast, if the schema already contains a Share objectclass then you could publish a prototype object using Extension to carry the new attributes, try it out and modify if necessary, and only then define a new class in the schema once you know it will work. One of the selling points for glue 2 is supposed to be that it's easily extendable to cover services beyond computing and storage, so we shouldn't put unnecessary obstacles in the way of that. Stephen -- Scanned by iCritical.