Second CGS-WG Meeting at GGF6
Minutes (draft, comments welcomed) taken by Larry Flon
 
Ellen reviewed the current JSIM UML model, and discussion followed around the model.
 
Q: shouldn't the model include job dependencies, e.g. this job requires this database system?
A: do we represent these dependencies within CIM or is this the responsibility of the scheduling system?
This remains an open issue, but there seems sentiment in the group to keep the scope of our work small 
and not address this piece right now.
 
Q: "job" could be a grid workflow; traditional batch job vs. grid job that runs on multiple computers.
A: Open issue, but again the sentiment seems to be to leave this for the future.
 
Discussion re: other WG's/RG's in the scheduling area.  In particular, the NPI WG has use cases for 
job submission that we should look at.
See http://www.pulsipher.org/npia/Artifacts/NPi.WP.RefModel.V1.0.pdf page 20ff.
 
It was deemed premature to review the JSIM document at this meeting.  The UML was enough to start with.
 
Discussion of the resource/execution environment as an abstract class, as suggested by Viktor 
in email. Left as an open issue.
 
Q: Should we take on the job of specifying grid services for job submission?
A: That is not within our charter.  That could be a new WG.
 
Discussion about using mail list for document review and updates, and creating a bugzilla database to track 
issues, like the OGSI-WG.  Larry will pursue that with GGF.
 
Discussion about where to scope the effort of the group in terms of schema delivery.  While we could deliver MOF 
as well as XSD and LDIF representations, it seemed to be the consensus that we should just do the MOF and 
leave the other translations as work for another WG.