Parrallel Forks and Parallel Processing

[Michael Pokraka]

Hi Alon,
I agree with you on the wait step - it also consumes unnecessary resource=
s.=20
Two thoughts spring to mind:
 
- How come the branch that 'does some processing' completes before the fi=
rst
one establishes it's wait status? OK, it's never guaranteed that less wor=
k
should arrive earlier but still it's a bit odd. Unless both flows do some
processing? In which case the design might profit from some more thought =
-
like how about putting the wait step into a new fork. I've had similar is=
sues
and managed to solve them by 'forking around' a bit.
 
- If you're using the event queue, try setting the instance linkage to qu=
eue
the event. This will incur an automatic delay without extra coding.
 
- If the 'does some processing' part is custom, maybe a WAIT statement in=
 the
code is less conspicuous than a wait step in the workflow (clutching at
straws)
 
OK, those were three thoughts, so I was mistaken...
Cheers
Mike
 
 
Alon Raskin wrote:
> Hi All,
>
> We have a situation where a workflow calls two sub-workflows in a paral=
lel
fork. One of the sub-workflows does some processing and raises an event .
The
> second sub-workflow needs to trap that event and handle appropriately. =
It
seems that the sencond sub-workflow is missing the event being raised
because
> the first sub-workflow raises it before the second sub-workflow has a c=
hance
to reach the 'wait for event' step.
>
> I know the obvious solution is to put a wait step for a few seconds to =
make
sure that the listening sub-workflow is ready but I hate that approach
because
> it feels like the 'wait step' is required due to technical limitations
rather
> then a business requirements. I know I could also raise an event when t=
he
listening sub-workflow is ready and only then process the first sub-workf=
low
but again this seems like modelling an event for the sake of the technolo=
gy
rather then the business process.
>
> Has anyone else experienced this issue? I would appreciate your thought=
s and
possible solutions.
>
> Regards,
>
> Alon
>
>