RE: Summing up on visibility(?) (really, I mean it this time 8-)

> -----Original Message-----
> From: Mark Baker [mailto:distobj@acm.org]
> Sent: Friday, January 10, 2003 11:19 AM
> To: Miles Sabin
> Cc: www-ws-arch@w3.org
> Subject: Re: Summing up on visibility(?) (really, I mean it this time
> 8-)
> 
> 
>
> 
> "The trade-off, though, is that a uniform interface degrades 
> efficiency, since information is transferred in a standardized form 
> rather than one which is specific to an application's needs."

I'm thinking more and more that this is the starting point for *all* web
services and not a differentiator between REST and SOAP/WSDL.  I'm further
inclined to say that this discussion has helped clarify how one can move
toward this by either making the information needed by applications other
than the intended one (e.g. intermediaries) available in at least two ways:
make the message content opaque and put the needed information in the
protocol headers (HTTP is particularly powerful in this respect), and/or to
make the message content "visible" using XML and related standards,
especially SOAP and its header extensibility model, WSDL to describe the
contents of the message body in a machine-friendly way, and numerous
emerging standard SOAP headers to support encryption, authentication,
signing, routing and reliable messaging across protocols and intermediaries,
choreography of multi-part messages, etc. etc. etc. 

For whatever it's worth, this discussion has deepened my understanding of
the traditional Web intermediary model based on ports and HTTP header
information, but it has INCREASED my appreciation for the value that XML and
SOAP add to that.  Just as the synergy of URIs, HTTP, and HTML created a Web
that is much more than the sum of its parts, the Web + XML have synergies
that are just becoming apparent ... and go far beyond the "XML is just
another opaque data format" perspective that it seems that REST advocates
prefer.  

The distributed object model without XML would not work well over the
internet, I do fully agree. But XML breaks out of the boundaries to achieve
RESTful *objectives* in a way that the REST "dogma" doesn't seem to
appreciate.  "Visibility" is an important principle, and XML makes data
visible beyond HTTP's dreams.  Likewise, I'm coming to appreciate the
uniform interface constraint for *data transfer*: a world of hard-coded RPC
methods in opaque messages would be as problematic as Mark claims, but
XML-serialized object interchanges can leverage both the uniformity of the
data interchange methods and the specificity of traditional distributed
object systems.  Whatever the theoretical rough spots here, they seem to be
increasingly overwhelmed by the network effect (e.g. all those people who
understand SOAP, WSDL, XPath, etc. are figuring out pragmatic workarounds).


> It seems that only Walden Matthews and James Snell[1] agreed with my
> characterization of visibility, of those who spoke up.  Is that enough
> to get it in to the WSA doc?

Is there some draft text that doesn't talk about cows or circles that you
suggest we put in?  Where would you suggest putting it? Can you live with
something like my summary above of the value of "visibility" and how
XML-based technologies complement the HTTP-based technologies? [I didn't
think so :-) ]

Received on Friday, 10 January 2003 12:34:53 UTC