SOAP interop

[Thomas Uram]

We have been working to document datatypes in the design documents.  The 
documents do need to be updated, so maybe the datatype definitions 
appear to be missing.

Bob:  Are you suggesting we write serializers now, to use until pyGlobus 
delivers WSDL support?   The schedule feels tight to me already, but I'm 
interested in your answer.

Tom


Robert Olson wrote:

> At 02:34 PM 1/30/2003 -0600, Ivan R. Judson wrote:
>
>> > Using schemas and wsdl don't get around the problem of having
>> > to clearly
>> > define the data types we are passing; it just changes the way
>> > they are
>> > described.
>>
>> Agreed, but we don't have to write serializers if someone else (eg,
>> pyglobus) is already working on them.  It's just a matter of 
>> understanding
>> how to use what they have, according to them.  Our data structures *are*
>> well understood, our choice is to not bother writing serializers when 
>> we're
>> planning on using wsdl/schemas, hence the pyGlobus serializers anyhow.
>
>
> Right. I'm arguing that we document, in terms of the structs that are 
> actually on the wire, the data being passed around.
>
>> The interop question is more for pyGlobus than us, since we're just 
>> writing
>> stuff over pyGLobus web services -- our stuff is transport 
>> independent (if
>> only logicallly).
>
>
> Hm. Stating "if only logically" doesn't help the person trying to 
> write a service that talks to the venue...
>
> If we are arguing that we have a web-service based system that is 
> interoperable with other web service-based tools, I think we need to 
> be able to back that up with examples (that we get working via 
> whatever means required). If we want to beat up the pyGlobus folks 
> about it, that's fine, but I think we need to do some diligence on the 
> problem.
>
> --bob
>