From: Andre Merzky <andre@merzky.net> To: Tom Goodale <goodale@cct.lsu.edu> Cc: Simple API for Grid Applications WG <saga-rg@ggf.org> Subject: Re: [saga-rg] wait on tasks... I think it would be nice to avoid the need for try/catch for wait, at least for the situation when the task finishes peacefully (enters Finished state). I agree that a concurrent cancel from the same application should be catchable, as that indicates a flaw in application logic (cancel in one thread, wait in another). So your proposed policy of allowing a wait to simply return success as long as the parameters are in a usable state, or to simply fails as long as the task might get into Finished state, would be a good solution. Andre. Quoting [Tom Goodale] (Aug 25 2005):
On Thu, 25 Aug 2005, Andre Merzky wrote:
Quoting [Tom Goodale] (Aug 25 2005):
On Wed, 17 Aug 2005, Andre Merzky wrote:
Hi all,
the wait currently throws a NotRunning exception if the tast is not running anymore. That basically mean that _every_ wait has to be put into a try/catch, since the programmer does not know in advance if the task is still active.
The can do a getState on the task.
You have a race condition then:
state = task.getState ();
# task finishes now
if ( state == Running ) { task.wait (-1.0); # throws an NotRunning exception }
So that does not help.
Good point, but it does if we also change the exception, although they still might want to put a try around it to catch someone cancelling the task in the meantime; on the other hand since that could only happen from within the same program, it's probably something the programmer can decide if they need to do.
Tom
-- +-----------------------------------------------------------------+ | Andre Merzky | phon: +31 - 20 - 598 - 7759 | | Vrije Universiteit Amsterdam (VU) | fax : +31 - 20 - 598 - 7653 | | Dept. of Computer Science | mail: merzky@cs.vu.nl | | De Boelelaan 1083a | www: http://www.merzky.net | | 1081 HV Amsterdam, Netherlands | | +-----------------------------------------------------------------+