AW: Behaviour on faulty agent determination...

[Zmudzin,Tomasz,FRANKFURT,Extern LG-DM]

Hi Michael,
 
"Why doesn't the system produce an error?"
 
Because it's pretty damned stupid -- or, more formally: "because existence,
LOCKS & validity are unfortunately not checked by the agent determination
program".
 
There is some justification for this, but it rests on philosophical grounds
that do not have be accepted by everyone:
 
- workflows failing because some user is locked / delimited somehow resemble
a dog wagged by its tail
- agent determination can be one of the more performance-intensive workflow
functionalities, so loading it with even heavier checks would only make it
worse (and users are supposed to be locked / invalidated rather than removed
anyway)
- there is a standard "Workitems without agents" report that should be run
-- which is softer than having to restart WFs after error,
 
so in general it is a matter of style.
 
As of the chocolate -- I'm too fat already, so let's do that for free...
 
Best regards,
Tomasz
 
-----Ursprungliche Nachricht-----
Von: Michael Klika [mailto:M.Klika at ids-scheer.de]
Gesendet: Dienstag, 12. August 2003 22:25
An: SAP-WUG at MITVMA.MIT.EDU
Betreff: Behaviour on faulty agent determination...
 
 
Hi There
 
I'm wondering about the following: If I provide a non existing user id as an
agent for a dialog task, than the list of users at the respective workitem
at runtime is simply empty.
 
Yes, of course I know that I have to check validity and existence of users!
But, just for curiosity, please tell me: Why doesn't the system produce an
error?
 
I wouldn't go so far to give out a piece of chocolate, but I would be
greatful for an answer :-))
 
Thanks, Michael