Addresses, identifiers, and cookie semantics

Chris,

On Thu, Nov 25, 2004 at 08:00:06AM -0500, Christopher B Ferris wrote:
> Mark,
> 
> Does this address identify me?
> 
> Occupant
> 57 Kerry Lane
> Whitinsville, Ma 01588-0366
> 
> No, it does not. It is an address. Nothing more than an address. It could 
> equally apply to me, my wife, my
> son or daughter or the people who move in after we sell the house.

It is an indirect identifier for you, as well as your family, or any
past or future party which might live at that address.  More direct
would be "Christopher B Ferris, 57 Kerry Lane ...".

"Addresses", in the common use of the term, are a special class of
identifier which contain information sufficient to enable interaction
(e.g. send a letter to you), in much the same way that some URIs are
"URLs" and enable the submission of a document to you, or the retrieval
of a representation of you.

> As for whether I meant ref params, no, I meant ref props.

As I pointed out, your use of RefProps for cookie-like semantics is
counter to the WS-Addressing specification.  I agree there's some
ambiguity there though, since, as most Web app developers know, you have
a choice between placing the session identifier in the URI or the
cookie, and each choice embodies different trade-offs.

> Many use cookies 
> to identify the session
> to which the exchange belongs.

Agreed.

> Yeah, yeah... I know, I know, but it hasn't 
> broken the web yet.
>
> I agree that reasonable people can disagree, but that in fact is all the 
> more reason to 
> permit both styles of interaction.

I agree.  That's why I'm not arguing for the removal of RefParams.

Just curious, but if you use RefProps as a cookie-like (contextual
state) mechanism, what would you use RefParams for?

Mark.
--
Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca

Received on Thursday, 25 November 2004 14:29:50 UTC