W3C home > Mailing lists > Public > public-sws-ig@w3.org > December 2004

Doubt about how to represent an OR condition in Result

From: Drew McDermott <drew.mcdermott@yale.edu>
Date: Thu, 16 Dec 2004 14:22:24 -0500
Message-ID: <16833.57456.202841.770887@dhcp130132249245.its.yale.edu>
To: public-sws-ig@w3.org


 > [Tatiana Vieira]

 > composed process A
 >   composedOf Sequence
 >     process B
 >      composedOf Split
 >        process C
 >        process D
 >        process E
 >     process F
 >     
 >   This way, process B output is the conjunction of the output of 
 > processes C, D and E. So, if everything goes right in process B, it
 > means that all the 3 processes execute correctly and the output is
 > Ok. However, if one or more of these 3 processes did not execute
 > correctly, also didn't process B. I don't know how to represent the
 > negative result, I mean, the fact that if one of the 3 processes
 > go wrong, does process B. The positive result condition is the
 > conjunction of the 3 positive results, and what is the negative
 > result condition, if I can't specify 3 different conditions for
 > exactly the same effect and output?

Right now Owl-S embodies a minimalist concept of process failure.  A
process fails if it doesn't complete.  That doesn't really fit with
the intuitive notion, where a process might complete but return an
error code; or it might signal an exception.

Anyway, a Split+Join fails to complete if and only if some component
fails to complete, so the semantics is clear enough.  It's probably
not what you want.

I believe Split differs from Split+Join in that Split succeeds as soon
as a component is started.  That means that a Split can never fail. I
don't really know what someone would do with this.

I think Owl-S needs a lot more work in this area.  Suggestions are
welcome.  

-- 

                                         -- Drew McDermott
                                            Yale University
                                            Computer Science Department
Received on Thursday, 16 December 2004 19:21:25 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 16 March 2008 00:10:59 GMT