Dear Norman Paton, I thank you for your proposal about an interoperability testing activity. Our first step projet aims to supply some data access tools and services on grids based on DAIS specs and WSRF.NET implementation. As soon as we reach an enough stable state for our development, we will highly consider your proposal and contact you back. I will also take a deeper look into the multiple web services interfaces design you mentioned about the OGSA-DAI project. I am absolutely convinced you are right about the interest of such an adoption. Time missed till now for such a various and efficient approach; we will attempt to solve it. Thanks again for your advices. With regards, Ny Haingo. Le 06/10/2006 12:12, Norman Paton a écrit :
Hi,
It's interesting to hear that you are working on an implementation of the relational realisation. Would you be interested in participating in an interoperability testing activity for the specification? Such an activity is likely to follow the approach outlined for the XML Realisation in:
http://www.ggf.org/documents/GFD.77.pdf
There is another implementation being developed by the OGSA-DAI project at EPCC in Edinburgh, and there was an earlier implementation developed at IBM Hursley.
With respect to WSRF, the evolution of standards for representing state in web services is outlined in:
http://download.boulder.ibm.com/ibmdl/pub/software/dw/webservices/Harmonizat...
The basic lesson (from my perspective) is that this is a space in which agreement has been slow to emerge, and that as a result one has to conduct specification, software design and implementation activities on the assumption that current proposals will be superseded. The WS-DAI specifications were designed to be conservative in their use of WSRF. The result of this conservative stance is that use of WSRF is optional in the specifications, and where WSRF is used, this makes little difference to the functionality of the service provided.
I don't think it's easy to provide advice for ongoing development projects that are currently using WSRF as to how to proceed, but all should proceed on the basis that WSRF will not be around for ever. As Malcolm describes, the OGSA-DAI team has architected their software in such a way as to allow multiple web service interfaces to be developed relatively easily in a thin layer. If you do use WSRF, you should certainly consider a similar approach. However, this approach helps the providers of a service and not the users of a service to cope with change. In the WS-DAI specifications, abstract names are mandatory in message bodies; this means that message bodies are the same whether or not you use WSRF. This looks like quite a good design decision with hindsight!
By contrast, I think it is easy to provide advice to standards activities as to what use they should make of WSRF or its emerging descendents! ;-)
Regards, Norman