[Biojava-l] GO web service

Dickson, Mike mdickson@netgenics.com
Tue, 3 Sep 2002 02:58:46 -0400


  I've been following the OGSA stuff with some interest.  And yes of course,
you are correct that you can build stateful services using the current
infrastructure.  What is more difficult to do is to return objects by
reference (quite easy in CORBA and RMI).  Even this is possible but you end
up rolling a lot of your own infrastructure.

  I feel pretty safe in saying that the large majority of web services
designs being developed today are stateless and service-oriented. This is in
contrast to distributed object programming using techniques like RMI and
CORBA.  IMO both approaches have merits and as you indicate research is
underway to extend the web services concepts to adopt some of the productive
aspects of the distributed object approach.

  So, let me restate: the challenge to a web services front end to biojava
is to develop definitions of service interfaces to mediate access to the
biojava data and program interfaces.  IMO this should be done using the
JAX-RPC based approach starting from Java interface and class defintions for
exported objects.  The alternative is to use a JAXM messaging style approach
which, while probably more flexible, will be harder for many to use because
of the need to handle formulation and conversion from/to XML documents in
the messages.


> -----Original Message-----
> From: Tom Oinn [mailto:tmo@ebi.ac.uk]
> Sent: Saturday, August 31, 2002 10:48 AM
> To: biojava-l@biojava.org
> Subject: Re: [Biojava-l] GO web service
> Dickson, Mike wrote:
> >Lastly of course
> >there is the stateless nature of web services vs. stateful behavior in
> >BioJava (and CORBA).  What does a good stateless interface to the stuff
> in
> >BioJava look like?
> Web services are not stateless by definition. Firstly, you can declare
> services as being session scoped, and therefore maintain state within
> them, and secondly there are a significant number of projects underway
> already to provide stateful service factory type implementations. I
> point you in the general direction of globus 3.0, or the Open Grid
> Services Architecture for more information. We are building a general
> eScience grid system based around this architecture, we have GO web
> services already running as part of the prototype for this system. See
> http://www.gridforum.org, http://www.ebi.ac.uk/escience and
> http://www.mygrid.org.uk for more information :)
> Cheers,
> Tom Oinn
> _______________________________________________
> Biojava-l mailing list  -  Biojava-l@biojava.org
> http://biojava.org/mailman/listinfo/biojava-l