- From: Jeffrey Mogul <mogul@pa.dec.com>
- Date: Tue, 17 Oct 95 11:46:20 MDT
- To: "Eric W. Sink" <eric@rafiki.spyglass.com>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
> I argue, therefore, that the Keep-Alive header conveys no useful
> information and ought to be dropped.
While agreeing with the rest of your message, I disagree here.
Providing a place for these kinds of parameters is the right idea
since we *will* run across a need for one, and that kind of
information should stay out of the Connection header itself.
Huh? To me, it doesn't make sense to add a header to the HTTP
spec whose only function is "to be determined". Either we need
it and its function should be specified, or we aren't sure and
so we shouldn't put it in the spec.
If some day you do find a need for some parameters related to
the persistent connection, and you want to experiment with this
before a subsequent version of the HTTP spec is written, you can
introduce a new header. To my mind, this is no different than
introducing a new parameter to use inside of a "Keep-Alive:"
header; either way, you get absolute freedom to define what
it is named and what it means, and no guarantee that anyone
else will agree with you about name and meaning.
The only reason to introduce a specific header now (in HTTP 1.1)
would be if it had to be protected from naive proxy behavior.
That is, "this header should not be forwarded" or "this header
must be rewritten when forwarding." But I have not seen any
proposals for such protection, and I can't think of any reason
why this would actually be needed here.
-Jeff
Received on Tuesday, 17 October 1995 12:18:59 UTC