[GRAAP-WG] Problems of WS-Agreement Specification

Karl Czajkowski karlcz at univa.com
Wed Sep 27 09:57:49 CDT 2006


On Sep 27, Tianchao Li modulated:
...
> 2. Agreement Template Property of Agreement Factory and Pending
> Agreement Factory Service
> 
> The agreement factory has a property of agreement template. However,
> the agreement template retrieval should be regarded as a part of the
> resource negotiation process to accommodate the different modes of
> negotiation.  For example, the  service provider might post their
> agreement template in a category (index) service as advertisement.
> 
> This might also cause security problems. For example, the service
> provider might want to provide different templates for different
> clients, or restrict certain clients to access template at all.
> 
> recommendation:
> 
> do not define the agreement template property in agreement factory and
> pending agreement factory service.
> 

I disagree with this recommendation. I think there are several less
invasive alternatives:

   1. an implementation is free to apply authorization checks to 
      control what templates are viewed by what clients during RP query

   2. an implementation is free to only list a limited set of "public" 
      templates in the PR, and provide an extended retrievel mechanism
      to get the "private" templates, as part of an extended negotiation
      system, etc.

   3. the process by which templates might find their way into indexes
      or registries is not defined nor restricted by the spec.

...
> We have different choices:
> 
> (1) merge agreement and agreement status port type
> 
> (2) modify AgreementFactory and PendingAgreementFactory portType
> 
> (3) specify the relation more explicitly in the documentation
> 

I am surprised by this. I think the specification should be stating
(3) that the result of our factory is a resource which implements
both. The reason for separating them is to allow someone else to reuse
just portions of the protocol, I think.  (But perhaps I have missed
some other discussion on this...)


karl

-- 
Karl Czajkowski
karlcz at univa.com



More information about the graap-wg mailing list