[drmaa-wg] Feb 21st 2006 Meeting Minutes

[Rajic, Hrabri]

Dan,

I do not think this is a good idea (in Tracker 1568 text):

"The spec should also state
 that calling drmaa_wait(DRMAA_JOB_IDS_SESSION_ANY) when the
 session contains no jobs will result in the call blocking until
 a job completes (necessarily after being submitted)."

In a single threaded app drmaa_wait(DRMAA_JOB_IDS_SESSION_ANY) blocks
forever.  Despite a bad app logic, there is no reason to be so strict.
If there are no wait able jobs it would be better to return an error.


Moreover Tracker 1400 resolution is incompatible with the above text:
"If there are no running of (**should be "or" here**) completed jobs the
routine returns DRMAA_ERRNO_INVALID_JOB error."


drmaa_synchronize() with DRMAA_JOB_IDS_SESSION_ANY could be used in use
cases where we need the fastest obtainable result to move to the next
stage.

Regards

	Hrabri


>-----Original Message-----
>From: owner-drmaa-wg at ggf.org [mailto:owner-drmaa-wg at ggf.org] On Behalf
Of
>Daniel Templeton
>Sent: Tuesday, February 21, 2006 1:19 PM
>To: Peter Troeger
>Cc: DRMAA Working Group
>Subject: Re: [drmaa-wg] Feb 21st 2006 Meeting Minutes
>
>I have added a comment to 1568 to clarify.  Let me know if I need to be
>clearer or if anyone disagrees.
>
>As an aside, is it possible to call drmaa_synchronize() with
>DRMAA_JOB_IDS_SESSION_ANY?  I don't think we foresaw that case in the
>SGE implementation.
>
>Daniel
>
>Peter Troeger wrote On 02/21/06 19:09,:
>
>>Dear all,
>>
>>the results of the phone conference are now reflected in the GridForge
>>tracker items, by comment and status changes. There is only one
pending
>>issue left, which needs some additional note from Dan in order to get
a
>>final solution.
>>
>>Best regards,
>>Peter.
>>
>>
>>Daniel Templeton schrieb:
>>
>>
>>>o Minutes for Jan 24th
>>>  - Accepted
>>>o Doc schedule
>>>  - Ready by April to submit everything
>>>o IDL doc strategy
>>>  - We have a few months (~6) before we're done with 1.0 spec
>>>  - Could use that time to start on IDL doc
>>>  - After we're done, we have to dissolve
>>>  - Put together IDL docs and submit them as proposed recommendation
>>>  - Java and .Net docs should also get submitted
>>>  - Be in better position for DRMAA 2 WG
>>>  - May need to recruit some new members if we want a new WG
>>>  - If our only goal is to have it published somewhere, GFD might be
a
>>>better path
>>>  - Peter and Dan will be in charge of getting the IDL doc published,
>>>somehow
>>>  - We just need to make sure we don't lose the work when we disband
>>>o DRMAA Wiki
>>>  - GridForge will be getting a Wiki
>>>  - Need to consider backups & maintentance
>>>  - Happy to have our own Wiki
>>>  - A lot of work to join an existing community
>>>  - Be better to leverage product communities than GGF community
>>>    . Where the momentum comes from
>>>  - We have a functional solution, so let's just keep things as they
are
>>>  - Backups are currently a worry, but there's so little to backup,
it's
>>>not much of an issue
>>>o Trackers
>>>  - Finally caught up
>>>  - All fixed except 1151
>>>    . Incompatible changes are problematic
>>>    . Delay until after 1.0 -- agreed
>>>  - Someone please close the fixed issues
>>>  - Will send around latest version of the spec
>>>  - 781 -- Will back-link to job category and native spec chapters
>>>  - 1125 -- Will borrow the text from the IDL doc and the SGE man
pages
>>>  - 1289 -- Will borrow the text from the IDL doc
>>>  - 1396 -- Delay for IDL spec
>>>  - 1566 -- Delay for post-1.0 discussion
>>>  - 1568 -- Handle offline
>>>  - 1587 -- Delay for IDL spec
>>>  - 1683 -- Will borrow the text from the IDL doc
>>>  - 1684 -- Rejected
>>>  - 1686 -- Agreed, need to augment for empty session (1568)
>>>  - 1727 -- Delay for post-1.0 discussion
>>>  - All fixed trackers must be worked into the spec
>>>  - Need independent verification
>>>    . Except for very minor fixes
>>>  - No remaining issues
>>>o SGE Exp Doc
>>>  - Still need to wrap it up
>>>  - Should be done by mid March
>>>
>>>
>>>
>>
>>
>>
>
>--
>******************************************************
>*         Daniel Templeton   UMPK18 x83749           *
>*        Staff Engineer, Sun N1 Grid Engine          *
>******************************************************
>* "What's the sense in never thinkin' 'bout the tomb *
>*  When you're much too busy returning to the womb?" *
>*                  -They Might Be Giants             *
>******************************************************