- From: Roy T. Fielding <fielding@avron.ICS.UCI.EDU>
- Date: Wed, 21 Dec 1994 02:50:18 -0800
- To: John Franks <john@math.nwu.edu>
- Cc: robm@neon.mcom.com, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
John Franks writes:
> According to Roy T. Fielding:
>>
>> Actually, I have not yet proposed any meaningful semantics for a SESSION
>> method other than it cannot be forwarded.
>
> I noticed. This makes it difficult to compare to an MGET proposal. I am
> eagerly awaiting some description of how SESSION and Connection headers are
> going to achieve the same user perceived performance as multiple connections.
Quite frankly, I don't give a damn at the present time whether SESSION
and/or "Connection: keep-alive" achieve the same user perceived performance
as multiple connections -- that has absolutely nothing to do with why they
must be recognized by 1.0 proxies, and thus why they should be in the 1.0 spec.
That is the only discussion I am interested in at the moment, because
I have no time right now to rehash all the discussions that were held in
the meetings and hallways of the IETF -- maybe later, but not until the
1.0 spec is reasonably stable and my mail inbox becomes livable.
The reality is that there must be some way to enable in-band communication
with an agent's nearest-neighbor in the communication chain, while at the
same time being reasonably assured that the communication is not mistakenly
forwarded down the communication chain. The question is NOT why we need
that communication, or what the communication will entail, but what
mechanism will be used to transport it and in which version of the protocol
(1.0 or 1.1) it should first appear.
......Roy Fielding ICS Grad Student, University of California, Irvine USA
<fielding@ics.uci.edu>
<URL:http://www.ics.uci.edu/dir/grad/Software/fielding>
Received on Wednesday, 21 December 1994 02:52:35 UTC