[kepler-dev] complex data types and web service harvester..

Ilkay Altintas altintas at sdsc.edu
Thu Nov 4 09:46:30 PST 2004


Bertram,

WSHarvester in the Alpha3 doesn't work due to some problem with the 
annotation. Shawn and I have been working on it since then. It is 
almost working again including the dynamic registy of services into the 
actor library.

For the complex type, web service features in Kepler has not been 
designed to handle complex Java types because of type mapping issues.  
It was a design decision that was made when this actor was design. So 
it is a feature not a bug. =) There might be a way to map complex 
XMLSchema types but I couldn't yet come up with a method that maps any 
XMLSchema type to a Ptolemy type. Any suggestions on this are welcome.

I know Jesus et al. had the same problem in the bio. workbench and they 
decided to build static clients for their web services instead of a 
dynamic proxy. We can also try similar approach if there are web 
services that require special type handling.

If I remember right from the Seek meeting in may, the web services from 
Kansas had java object types. So it might not be possible to use the 
web service actor as a client for them.

Ilkay




On Nov 4, 2004, at 7:42 AM, Bertram Ludaescher wrote:

>
> Ilkay et al:
>
> I was just trying to  harvest the WSDL on this page:
>    http://seek.ecoinformatics.org/Wiki.jsp?page=TaxonMtg_Nov2004
> but didn't succeed.
>
> I think there might be  several reasons why:
>
> (1) I don't know how to actually use the web service harvester ;-) Can
> we put some info under right-click."Get Documentation"?
>
> (2) I'm using the alpha3 so maybe the ws harvester doesn't work there?
>
> (3) There seems to be a fundamental problem (Matt just pointed out
> during our SEEK mtg), i.e., that the ws actors/harvester cannot handle
> comples WSDL/XML Schema types.
>
> I suspect that other ws clients (e.g. Java based ones) might have
> similar problems in generating a good mapping from complex XML Schema
> types to a Java object type (or in our case a Ptolemy complex type).
>
> Any ideas anyone? (Is there already a bugzilla entry for this?)
>
> Bertram
>
> _______________________________________________
> kepler-dev mailing list
> kepler-dev at ecoinformatics.org
> http://www.ecoinformatics.org/mailman/listinfo/kepler-dev




More information about the Kepler-dev mailing list