Re: "Reliable" web services for Next Big Thing? (was RE: Agenda for 5 December WSA telcon)

On Thu, Dec 05, 2002 at 10:50:09AM -0800, Sandeep Kumar wrote:
> Mark,
> Could you elaborate as to why you would be against HTTPR?

Hi Sandeep.  There's really two main reasons, and either one by itself
would be enough to convince me to dislike HTTPR. 8-)

First, it treats HTTP as a transport protocol rather than an as an
application protocol.  Yes, I know I sound like a broken record with
that, but what can I say? 8-/  In the context of this discussion, let me
rephrase my position as this: HTTP defines a coordination language, and
treating it as a transport protocol removes all the yummy goodness of
the coordination semantics that exist over an HTTP connection.  A
proper reliable extension of HTTP would *extend* those coordination
semantics (and reuse others, like GET) with reliability features.

The second reason is just the same reason that people have been
chiming in with their experiences about; dealing with reliabile message
delivery at the transport layer (the *real* transport layer 8-) is very
often a mistake, as it hides important information from applications
that they really need to know in order to cope with inevitable network
failures over the Internet.  The Jim Waldo "Note on Distributed
Computing" paper describes this issue really well, IMO;

http://www.sun.com/research/techrep/1994/abstract-29.html

MB
-- 
Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca
Web architecture consulting, technical reports, evaluation & analysis