mpiexec SGE integration

[Pete Wyckoff]

raysonho at eseenet.com wrote on Fri, 11 Feb 2005 20:54 -0800:
> I think this is what needs to be done:
[..]

I think your list is right, and can't think of any missing parts.

> Pete, can you check if I missed anything?? Also, is there anything in the
> TM API that you think should be improved??

The TM API isn't very much fun.  Note especially the race conditions
associated with killing a task for which you are already waiting for an
obit message, in wait_tasks().  You don't have to plan for that up front
but some thought will have to happen later.

I would leave out the tm_obit() call and make that a mandatory part of
tm_spawn(), just like it is for tm_kill().  And make sure the server
always delivers both messages to avoid that race.  But don't fret about
that yet, it's a relatively minor detail.

Good luck with the coding!

		-- Pete