RE: Visibility

> -----Original Message-----
> From: David Orchard [mailto:dorchard@bea.com]
> Sent: Thursday, May 08, 2003 11:15 AM
> To: 'Mark Baker'; www-ws-arch@w3.org
> Subject: Visibility
> 
>
> >
> > Visibility redundant?  My goodness.  IMO, it's the single
> > most important
> > property than an Internet scale architecture can have.  Roy
> > just wrote a
> > bit about it, in fact;
> >
> > http://groups.yahoo.com/group/rest-discuss/message/3649


I don't have strong feelings on whether visibility is redundant or not, but
I think  that (as I sketeched out in the section 1.6 draft) that REST
doesn't have a monopoly on it, and any SOA that employs XML and especially
SOAP is going to reap most of these benefits.  Let's look at Roy's points:

"o efficient intermediaries"

We're seeing XML-aware "edge appliances" that can deal with XML
parsing/pattern matching at almost network speed; I don't think efficiency
is a profound issue at the architecture level.

"- security across trust boundaries"

There are a lot of XML/SOAP-based security technologies out there!

"- extensibility - 
Visibility calls for more explicit descriptions within messages, as
opposed to relying on external IDLs or XDR-style typing by schema,
which in turn makes it easier to extend the functionality later."

Right, and XPath is increasingly being used (at least in principle, but we
ARE talking about architecture here not deploying systems today) to do
exactly that.

"- evolvability "

I must admit I don't understand this point in either the REST or XML-powered
SOA context.

"- performance Visibility is necessary for cacheability."  

I'm not aware of XML-powered cacheing intermediaries, but they could
logically exist.

"o understandability
It is simply easier to build complex systems if you can see and
understand the individual interactions."

Sure, an oft-cited reason why text-based formats such as XML are more
understandable / debuggable than their binary predecessors.

"o reusability "

The network effect caused by the universal adoption of XML/SOAP generates
even more reusability than the HTTP headers!

Once again, Mark, you are welcome to bring this up with the Powers that Be,
but I honestly think we've got so many counter-arguments to every point you
could raise that it will not be a winning argument for you.... not to
mention the fact that there are more XML geeks than REST geeks in that
group!  I grepped for "XML" in Roy's thesis the other day and didn't find a
single instance; I simply don't think he considered (back then!) the
possibility that XML provides the same visibility benefits that HTTP headers
did in the days of opaque content.  If he has considered the visibility
issue since XML became ubiquitous, I would be very interested in what he has
to say.

Received on Thursday, 8 May 2003 11:50:19 UTC