RE: [classad-users] Newbie Question: Evaluation of logicalexpress ions

[Uwe Maurer <uma@xxxxxxxxxxx>]

Hi Alain,


> >Hmm. The reason for asking about shortcutting expression 
> evaluation is a
> >vague idea of extending the built-in functions of classad with some
> >application specific functions that are quite expensive  to execute
> >(database queries, RPC calls and the like).
> 
> That makes sense to me. I would have to think about the best 
> long-term 
> implementation, but either of the approaches you outlined 
> would seem to work.
> 
> I'm curious what you are using ClassAds for. If you don't 
> mind talking 
> about it, I would be interested in hearing about it.

Sure. Here we go: We working on a component based web application, where the
users are able to configure "their" application. This configuration is
represented as a finite state machine in our data model. As soon as
"something" happens (some/the user himself enters the his website for
example), the reaction of the application is determined by a transition of
the user specific state machine. The possible transitions are guarded by a
set of contstraints. That's were we need something like classads: The
current state ist represented as a set of name/value pairs (= a classad),
the constraints could be classad expressions. A transition is "valid"
(possible, allowed) if the state matches the contraints.

We need more functionality as I said above: We need to check database-stored
quotas, do CORBA calls to other components and the like. In most cases,
however, the transition is determined only by the type of event that arrived
in the current state. So we have conditions like

   "last_event == DATA_ENTERED && not quota_exceeded(USERID)"

"quota_exceeded" would be a quite expensive function, invoking a server via
CORBA, making it querying somethin from a database, compute and return the
result. So it is a requirement for us, that in cases where "last_event" ist
something different, the "quota_exceeded" is _not_ executed.

Of course we could implement our own (simple) contraint language - but why
not use something that's proven und nicely designed by others - like
classad, for example.


Regards,
Uwe
Condor Classads Info:
http://www.cs.wisc.edu/condor/classad/