[cgs-wg] Minutes of today's telecon

[Larry Flon <flon@ISI.EDU>]

will be posted on GridForge.  But they are embedded here in order to spark a
discussion on this list.  So please comment!

	Larry

Minutes of the 4-May-2004 CGS-WG Telecon

Attendees:
	Larry Flon, USC/ISI
	Viktor Mihailovski, IBM/Germany
	Brian Collins, IBM/UK

The first problem noted was that the conference call logistics were broken.
Once these three people were connected, no one else could connect.  We are
tracking down this problem and will have it fixed for the next call.

We began by reviewing the first draft UML model for a logical database
schema.  This model can be found on GridForge at 

https://forge.gridforum.org/projects/cgs-wg/document/SRIM_UML_Current_Draft/
en/1

The following questions arose:

o Should we represent both Tables and Views?

    The answer appears to be Yes, as they both are relevant to the use
cases.

o In the case of NamedColumnSet and QueryCOlumnSet, should the distinction
be maintained?

    There does not appear to be a reason to maintain this distinction. The
suggestion is to drop these two classes, and give the class ColumnSet a name
attribute.

o What about the classes to represent DataType?

    This is an open issue.  We are not sure at what level we should define
data type.  It is a Pandora's box if we don't take a simplistic approach, at
least initially.

In general, we felt that prioritization of the Use Case document is really
needed.  We can probably develop a logical database model that will support
Discovery, in short order.  But if we need to support Administration, then
the model is going to be much more complex.  Perhaps we should consider
focusing on Discovery for now?

We will continue this discussion on the mailing list.  There will be another
telecon next Tuesday, May 11.