[DRMAA-WG] Minutes from August 28th meeting

[Roger Brobst]

August 28th DRMAA-wg Meeting Minutes


Attendees:
    Hrabri Rajic, Andreas Haas, Peter Troeger, Dan Templeton, Roger Brobst

A. Meeting secretary for this meeting?
  + Roger Brobst

B. Acceptance of the last meeting minutes
  + July 31 minutes accepted without change request


The Language-Independent spec was mistakenly marked as a
new independent doc, not as a second stage GDF.22 doc update
Steven Newhouse has been contacted and has indicated that
the wg needs to add a few lines summarizing the changes
(this will not restart and timelines).

General consensus to add the requested information to LI spec.
AI: Hrabri has agreed to notify Steven that requesed
    changes will be added.
AI: Hrabri has volunteered to summarize the changes and
    and post summary for further discussion.


C. Presentation of updated IDL spec for OGF 21
  + Peter distributed link to updated IDL spec,
    which now contains more details about reestablished sessions
    and reaped jobs. Specifically, Section 9, page 23 has rearranged
    text from the LI spec; and jobProgramStatus is now jobStatus.
    There are no new semantics. Just more text, mostly copied.
    The document is in the "OGF21" folder.

AI: Provide Peter with feedback on above changes

  * Discussion about 5.16 InvalidJobException :
    The job specified by the given job id does not exist,
    or was already reaped by a call to Session::synchronize()
    with dispose==TRUE.

AI: Since the wait() description discusses reaping, it should
    also be part of the exception description.

AI: Dan to update Java binding spec with the jobStatus method
    renaming.

AI: Although spec enumerates drmaa calls which may precede
    drmaa_init(), it is desirable that the doc for the
    *_control and *_job_ps routines reiterate that drmaa_init()
    should be called previously.


Regarding OGF21 ..
    Dan is planning on attending at least part of the time.

    Peter has also requested a DRMAA session with 45min about
    the IDL document and the expereince reports. No answer so far.

    Rough consensus that that the OGF session would be a good
    opportunity to clarify our intention of publishing IDL and
    Java docs after the LI spec has been approved with the ADs
    (to avoid misunderstandings about the wg future).

AI: Be clear that we're continuing the DRMAA work with the IDL spec.


Regarding coordination of public comment period (DRMAA, Torque) ...
 +  Desirable to continue to gather feedback to ensure that
    we follow the intent of comment period.
    We have 2 months for the LI spec.

    Possible contributors: Posnan guys, SAGA folks, Sun customers