W3C home > Mailing lists > Public > public-sws-ig@w3.org > January 2005

RE: Question about the Congo process example

From: Collin Hsu <collin@w3china.org>
Date: Thu, 13 Jan 2005 00:47:48 +0800
To: "'Tatiana Vieira'" <tascvieira@yahoo.com.br>, "'Manshan Lin'" <lmshill@gmail.com>, <public-sws-ig@w3.org>
Message-ID: <E1CollV-0002qZ-9O@frink.w3.org>
 

I understood this. The problem for me is that in each result the effect is
defined, but it isn't defined the output (using the property withOutput).
Isn't it always necessary to specify the output of a process, even though
it's an atomic process?  
 
First, OWL-S permits the absense of "withOutput" with a Result, because
there is no cardinality restriction defined for the "withOutput" Property in
process.xml.
 
 
 Is it enough to specify the type of the output data (through the effect
condition)? 
 
Second, probably closer to your question, an OutputBinding ( range of
withOutput ) is used to specify cross-process data flow. Take the process
CreateAcctSequence as an example, it has an Output called CreateAcctOutput
whose value is bound with the Output of CreateAcct - a sub-process of
CreateAcctSequence. To describe this data flow, an OutputBinding is defined
in the Result of CreateAcctSequence. However, the Output of LocateBook does
not receive any data from other process, at least in the model presented in
the Congo example, so there is no such OutputBinding definition in the
Result of LocateBook. Normally only output of composite process will has a
OutputBinding definition. 
 
Hope this helps.
 
 
Collin






Received on Wednesday, 12 January 2005 16:54:37 GMT

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