After spending another night trying to figure out what we would like
to say in regard to RFC 3205, I am ready to declare this as a "waste
of time".  The basic problem is that the RFC covers a great deal of
topics in a mostly accurate fashion, and is directed towards misuse
of HTTP as a transport layer protocol.  As far as that is concerned,
there is only one appropriate answer: HTTP is an application protocol
and it makes no sense whatsoever to use it as a transport.  However,
unless the TAG is willing to make a definitive statement to that effect,
I am probably the wrong person to be writing a critique of 3205.

I see no point in addressing the individual problems that some W3C uses
of HTTP might have with the recommendations in RFC 3205.  3205 merely
describes common ways in which ignorance of HTTP practice will cause
problems when introducing protocols that abuse HTTP semantics.  Although
I would have written it differently (e.g., the recommendations in
section 8 are simply wrong -- HTTP semantics must not be hidden by
the application even when it is layered), I'd rather just inform
people how to use HTTP correctly.

Finally, I believe that the reason 3205 is a BCP instead of an
Informational document is so that it can be revised over time.
There are areas in which the document is already out of date
(it was written long before its publication date), such as the
discussion of "s" URLs and use of 40bit encryption.  It can be
revised by the IETF when (or if) it proves to be inacccurate in
regards to some deployed Internet protocols.


Received on Monday, 22 April 2002 05:40:03 UTC