Public Comment

[comment-form@oasis-open.org]

Comment from: abeti@cs.unibo.it

Name: Luca Abeti
Title: Student
Organization: University of Bologna
Regarding Specification: WS-Resource 

Object: State information in WSRF

Dear WSRF developers,

I'm a student at the University of Bologna (Italy) and I'm going to take my CS degree with 
a thesis in GRID and WSRF technologies.

I read the OASI specifications and articles, together with the GT4 tutorial and WSRF FAQs but 
I can't see the "state" boundaries yet...


THE QUESTION IS:

What are boundaries beetwen stateful content and application specific content?



I mean your definition for state "data values that persist across, and evolve as a result 
of, Web service interactions" is not explanatory of which data must be managed by a 
WS-Resource. 


Is it state content ONLY meta-information (e.g. GenericDiskType), support infomration 
(e.g. accumulator variable), session information (username, job) an so on?
or
also information concerned with the message exchange can be considered "stateful information" 
(e.g. a database row, a file in the filesystem, a meteorological sensor output, etc.)


I can't see a clear difference between this two conceptual information space in your 
Ws-Resource definition. Morover there are some situation where it is hard to distinguish if
an information must be carried in "state data" or in "Web Service exchanged data". 

In some presentation of Globus you give the following examples of WS-Resource: Files on a file server,
Row in a database, Jobs in a Job submission system, accounts in a bank (Ben Clifford, GGF Summer School 2004)
In the GT4 tutorial you make a clear distinction between "state data" (e.g. "accumulator") and data
exchanged (e.g. "parameters" of the add function).

For a file resource we can see clearly the general distinction between state (filesystem informations)
and non-state information (files themselves). However, it is not clear in a WS-Resource context how 
this information must be managed. 
Moreover in some application, as remotly managed instruments, the distinction is implicitly 
more complex (i.e. coordinate of position in a robotic hand, a meteorological data taken directly from 
a sensor and more) and my doubt increase.


Yours Faithfully
						Luca Abeti