Re: DR302 (decentralized extensibility)

On Wed, 15 Nov 2000, Glen Daniels wrote:

> Rekha:
>
> An example of decentralized extensibility without prior agreement might be
> as follows (folks who originally worded this, please correct me if I'm not
> in line with your thinking here):
>
> A client attempts to access service S.  While there is certainly some sort
> of a-priori knowledge which allows the client to know the sort of service S
> offers (let's say a tax-calculation service in this case), the client does
> NOT know that the company who provides S recently added an authentication
> extension AUTH to the service.  So the initial call fails with a fault
> indicating the missing authentication information is the culprit.  Luckily,
> the client has a repository of available extensions, one of which happens to
> be AUTH (the client version).  So it can then simply trigger the AUTH
> extension, which gathers authentication information from the user and
> includes that in a resent copy of the XP message, which succeeds this time.
>
> Service discovery would make this process quicker, but it isn't necessary in
> this case.

What Glen said, though with slightly stronger emphasis: it is really
pretty important to be able have safe encounters with unknown extensions,
and to make this dependent on service discovery machinery is risky.

These concerns were very well articulated in the old HTTP-NG effort, where
a  lot of attention went into avoiding the extensibility problems of HTTP:

http://www.w3.org/Protocols/HTTP-NG/ ->
http://www.w3.org/Protocols/HTTP-NG/1998/11/draft-frystyk-httpng-overview-00

                HTTP-NG Overview
                Problem Statement, Requirements, and Solution Outline
[...]
 2.2  Distributed Extensibility....................................4

       A wide range of applications have proposed various extensions to  HTTP
       including distributed authoring, collaboration, printing, and  remote
       procedure call mechanisms leading to a growing tension between
       dynamically extensible applications and public, static
       specifications...
	[...+a few more good paragraphs that might be pinched for XP?]


Dan



>

> --Glen
>
> ----- Original Message -----
> From: Rekha Nagarajan <rnagaraj@calico.com>
> To: <xml-dist-app@w3.org>
> Sent: Wednesday, November 15, 2000 12:02 AM
> Subject: DR302
>
>
> > DR302 states:
> >
> > "The XML Protocol MUST support modular extensibility between communicating
> > parties in a way that allows for decentralized extensibility without prior
> > agreement. The WG must demonstrate through use cases that the solution
> > supports decentralized extensibility in a modular and layered manner."
> >
> > We believe this is out of scope (the "without prior agreement" part),
> > because service discovery is out of scope, and we don't see how we can do
> > one without essentially solving the other.
> >
> > - Rekha
> >
> > Rekha Nagarajan
> > Calico Commerce
> >
>

Received on Wednesday, 15 November 2000 04:17:26 UTC