W3C home > Mailing lists > Public > www-ws-desc@w3.org > March 2002

RE: Reference requirements

From: Krishna Sankar <ksankar@cisco.com>
Date: Mon, 4 Mar 2002 11:09:21 -0800
To: <www-ws-desc@w3.org>
Message-ID: <NABBJDOPDKGCDCNBNEDOMENHHCAA.ksankar@cisco.com>
Paul,

 | I may not in my implementation keep an object alive for a week. But yes,
 | I want a logical object to be accessible to the client for a week. The
 | only way that can happen is if I give them a reference to the object.
 | Maybe the reference is implicit, like a UUID.
 |
 | >...
 | > [KL] Again, referencing an object and passing an object by
 | reference, in my
 | > opinion a two different thing. In the WSDL interface term, passing by
 | > reference is more relevant.
 |
 | If you pass a reference as a parameter is that not the same as passing
 | "by reference?"
 |
<KS>
	Here are some of my thoughts :

	1.	Re. the PO example, I assume you are talking about passing reference as
a parameter - for example a UUID or some form of conversation ID. Of course
this is domain specific (could be a string, an integer, a multi part complex
object, ...). IMHO is outside the WSDL scope as it is like any normal
parameter.

		On a related note, normally we keep it for a long time (or short time) -
thru some social and technology process i.e. get the number thru phone or
from the web, enter it in a database or keep it in a napkin or in the back
of an envelope. This includes seat numbers, reservation numbers, order
numbers, registration codes,  ...

	2.	The object references is a software artifact (as compared to a business
artifact which is what #1 is) and that could be in our domain to solve.

	3.	Another point that stands out is the TTL which needs to be defined,
monitored and handled

	4.	What about validation ? Most probably none, as we really cannot make any
intelligent conclusions (except when we dereference it)
</KS>

cheers

 | -----Original Message-----
 | From: www-ws-desc-request@w3.org [mailto:www-ws-desc-request@w3.org]On
 | Behalf Of Paul Prescod
 | Sent: Friday, March 01, 2002 3:29 PM
 | To: 'www-ws-desc@w3.org'
 | Subject: Re: Reference requirements
 |
 |
 | "Liu, Kevin" wrote:
 | >
 | > ...
 | >
 | > [KL] State management and stateful objects are two different
 | things. All
 | > applications need to manage state somehow, stateful object is
 | only one way
 | > to do that. Even in a 'classic' n-tier application, many
 | people consider
 | > stateful object as not scalable.
 |
 | You'll have to explain the difference to me. Remember that we are
 | talking about a network interface, not an implementation strategy. There
 | are no "objects". There are only resources in the Web sense or endpoints
 | in the SOAP sense.
 |
 | > In your example, Do you mean you want to keep your stateful
 | object alive for
 | > a week?
 |
 | I may not in my implementation keep an object alive for a week. But yes,
 | I want a logical object to be accessible to the client for a week. The
 | only way that can happen is if I give them a reference to the object.
 | Maybe the reference is implicit, like a UUID.
 |
 | >...
 | > [KL] Again, referencing an object and passing an object by
 | reference, in my
 | > opinion a two different thing. In the WSDL interface term, passing by
 | > reference is more relevant.
 |
 | If you pass a reference as a parameter is that not the same as passing
 | "by reference?"
 |
 | > > 3. As for WSDL, it already has some construct for expressing in/out
 | > > parameters using ParameterOrder attribute of Operation - any
 | needs to go
 | > > further?
 | >
 | > That's an unrelated feature.
 | > [KL] Can you explain?
 |
 | Perhaps you can. What do in/out parameters have to do with passing by
 | reference? Yes, in/out parameters are "like" passing by reference in
 | some programming languages but we're talking about networking here.
 | Passing by reference is not a hack to get around some programming
 | language's limitations on returning references.
 |
 |  Paul Prescod
 |
 |
Received on Monday, 4 March 2002 14:12:15 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:19 GMT