W3C home > Mailing lists > Public > www-ws@w3.org > November 2004

RE: Stateful Web Services...

From: Newcomer, Eric <Eric.Newcomer@iona.com>
Date: Tue, 2 Nov 2004 12:47:07 -0500
Message-ID: <C7F58A2037A37D4B9212D4F09589B3FA0162973E@amereast-ems2.boston.amer.iona.com>
To: "Francesco Furfari" <francesco.furfari@isti.cnr.it>, "Jim Webber" <Jim.Webber@newcastle.ac.uk>
Cc: <hf0722x@protecting.net>, <www-ws@w3.org>

Actually we are proposing a persistent context management standard in the WS-Context specification within the OASIS WS-Composite Application Framework.  It has just reached committe draft status.


-----Original Message-----
From: Francesco Furfari [mailto:francesco.furfari@isti.cnr.it]
Sent: Tuesday, November 02, 2004 12:29 PM
To: Jim Webber
Cc: hf0722x@protecting.net; www-ws@w3.org
Subject: Re: Stateful Web Services...

Jim Webber wrote:

>Hey Francesco,
>>So,  in my opinion, web services are simply managed software 
>>that expose service interfaces in a standard 
>>machine-processable format and we could  improve 
>>interoperability if the standardization process concerns the 
>>state rapresentation too.
>On the contrary you introduce an interoperability problem by exposing
>yet another set of actors that I have to interoperate with.
mmm... I think I have to read a lot of things before to say some one of 
useful  ;-)
It seems to me that a lot of misunderstanding comes from different 
"stateful  web services" interpretation ...
I think to start from Roy Thomas Fielding's work that has already been 
cited  in this very long debate
to appreciate all the positions expressed here.

Some weeks ago on java.sun.com web site there was a nice graphical map 
of all Java related technologies ...
I think a similar "WSMap" could be very useful to understand all 
different movements around Web Services
with links pointing to the right resources ... with wiki technologies 
should be enough easy to set up it ...
any volunteer ?

anyway ...yes it is , I introduce a new problem because it wasn't 
addressed neither solved before (I suppose).
Yes, may be in such way we add another layer on top of the protocol 
stack, with performance iussues (etc.),
but it could be convenient if  we hide this complexity inside the WS 
middleware. From the developer
pointview we shouldn't  care more about how the session or context  is 
managed inside a
specific "developer kit \ container" (at least we do it in a standard 
way). From machien to machine
perspective we could move the state (or references) between differents host.

All this requirements can be considered  suitable for all domains?
Is there a large demand to justify the effort?
Is it valid only for a limited numebr of applications?

At the moment others have better answers than me ;-)

Received on Tuesday, 2 November 2004 17:47:35 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:37:11 UTC