- From: Bijan Parsia <bparsia@cs.man.ac.uk>
- Date: Sun, 11 Jun 2006 13:14:32 -0400
- To: Jacek Kopecky <jacek.kopecky@deri.org>
- Cc: Jim Hendler <hendler@cs.umd.edu>, SAWSDL public list <public-ws-semann@w3.org>
On Jun 10, 2006, at 3:00 PM, Jacek Kopecky wrote: > Dear Jim, > > I agree with you very much on the usefulness of knowing whether an > operation has some significant side effects or not. It seems that you > basically need to know whether an operation is "safe" in Web > Architecture terms [1]. Please note that WSDL 2.0 [2] has an extension > called "Operation Safety" which allows you to annotate your WSDL > with a > claim that an operation is safe. This mechanism is trivially > portable to > WSDL 1.1 as well. Do you think this would help you? That seems to cover a fair bit. > Secondly, I am of the opinion that preconditions and effects can > both be > referenced using a modelReference. Yes, you cannot just point to a > simple expression, you have to point to something that will say that > this expression is a precondition or an effect, but that also gives > you > the opportunity to specify exactly what you mean by a precondition > or by > an effect, i.e. what is the context for evaluating the expression. I would think you would need both the precondition and effect bundled together, since they often share variables. > This > is very similar to a proposal we recently had for moving interface > category to a tiny ontology + modelReference. Do you think that this > approach would be useful for you? > > Lastly, I am prepared to consider something like preconditions or > effects, whether or not it is based on modelReference, provided that > somebody gives the WG a fully-fledged proposal and it does not > generate > significant controversy. [snip] Ha. Hmm. Actually, one could do something reasonable I think with SPARQL (not *THE* P&E, but *an* P&E). Cheers, Bijan.
Received on Sunday, 11 June 2006 18:49:47 UTC