RE: TAG discussion of WS visibility issue

An interesting statement.

> I don't think previous attempts to deploy decentralized programmable APIs
have had sufficiently
> powerful tools available.  Effectively, WSA is betting that the XML and
URIs will be sufficient
> to permit the same levels of visibility that have previously required
generic interfaces.  Now, if
> we didn't have XML and it's self-describing nature, I'd be inclined to
agree with you that the effort
> would likely fail.  It all hinges on XML.

HTTP also bundles the semantics of that simple uniform interface, so does it
hinge on not only the
use of XML to describe the interface, but the use of XML to describe the
semantics of the interface
as well?

-bryan

-----Original Message-----
From: David Orchard [mailto:dorchard@bea.com] 
Sent: Tuesday, July 15, 2003 3:13 PM
To: 'Mark Baker'; www-ws-arch@w3.org
Subject: RE: TAG discussion of WS visibility issue



I think that there is a problem with the internet process for registering
applications as it exists right now because it doesn't scale to the number
of type of applications we want to deploy.  The process of registering for
tcp ports doesn't really lead to the flexibility and scale that we want.
With Web services, we can acheive a newer and decentralized way of deploying
applications.  The solution of making every application re-use the HTTP
rules for port 80 (or 443) is one solution, but seems to be too limiting.

I don't think previous attempts to deploy decentralized programmable APIs
have had sufficiently powerful tools available.  Effectively, WSA is betting
that the XML and URIs will be sufficient to permit the same levels of
visibility that have previously required generic interfaces.  Now, if we
didn't have XML and it's self-describing nature, I'd be inclined to agree
with you that the effort would likely fail.  It all hinges on XML.

> I know you agree that visibility is important for firewalls, so we 
> must also agree that there is some amount of visibility below
> which messages
> will not get past firewalls.  Empirical evidence gleaned from the
> existing Internet suggests that using a generic interface (not
> necessarily a uniform one) is necessary for use at Internet scale.
> Right?  If you disagree, can you name a single well-deployed system on
> the Internet today that isn't built with one?  This is a precarious
> position I'm taking, if all it takes is a single example to prove me
> wrong, no?  So name one, and I'll write an entry on my blog praising
> the value of object-specific interfaces, and will never speak of this
> issue again.
>
> > The main point I was trying to make in reproducing the
> quote is the idea
> > that this is all about engineering tradeoffs
>
> Of course.
>
> > and not incommensurable
> > paradigms.
>
> If an engineering tradeoff is made that prevents requirements from 
> being met, then the system can fail if those requirements were 
> sufficiently important to the success of that system.
>
> MB
> --
> Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca
>
>

Received on Thursday, 17 July 2003 15:44:56 UTC