Begin forwarded message:
Resent-Date: Mon, 25 Mar 1996 19:06:24 GMT
To: Harald.T.Alvestrand@uninett.no
Cc: "Roy T. Fielding" <fielding@avron.ics.uci.edu>,
http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Subject: Re: About that Host: header....
In-Reply-To: Your message of "Mon, 25 Mar 96 13:00:56 +0100."
<199603251201.AA141525267@hplb.hpl.hp.com>
Date: Mon, 25 Mar 96 14:06:05 -0500
From: jg@w3.org
X-Mts: smtp
Resent-From: http-wg@cuckoo.hpl.hp.com
X-Mailing-List: <http-wg@cuckoo.hpl.hp.com> archive/latest/81
X-Loop: http-wg@cuckoo.hpl.hp.com
Resent-Sender: http-wg-request@cuckoo.hpl.hp.com
Strategies that are 4 pages long are usually incomprehensible.
Here is my personal belief of a reasonable evolutionary strategy for
HTTP (but in one page):
HTTP 1.0. Current deployed protocol
HTTP 1.1 Has host header, etc, given the current consensus on
the mailing list. Current list of other stuff that makes 1.1.
HTTP 1.2?
HTTP 2.X Multiplexing protocol, still MIME based.
A number of us have been working on such a thing for the last few
months. See http://www.w3.org/pub/WWW/Protocols/HTTP-NG/mux/WD-mux-960315.html
for a multiplexing protocol design (a more compact and extensible version
of Simon Spero's SCP).
(Note that this is a draft of a draft and is guaranteed to change, rather
than an actual W3 working draft). Also note that this MUX protocol
can multiplex almost anything, so would likely be used with 3.X.
We're prototyping implementations of this now at W3C to get some
experience with it, and Paul Leach is threatening an implementation as well
in his research server. The big issue here is not
the mux protocol itself (though I'm sure you can find ways to improve it :-)).
but as usual, the transition strategy (how to know when you can
use the new, improved protocol).
Depending on timing of the multiplexing protocol, one might either do
a version 1.2 or go directly to a multiplexing protocol.
HTTP 3.X complete redesign, once we really understand what we're doing with the
web. Likely a binary protocol of some sort. A.K.A. HTTP-NG. I think this
will take longer to do than we can wait for multiplexing, so think that
the multiplexing intermediate step is worthwhile.
- Jim