W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > May to August 1996

Re: Rev81: COMMENT: 5.2 The Resource Identified by a Request

From: Shel Kaphan <sjk@amazon.com>
Date: Mon, 3 Jun 1996 12:41:27 -0700
Message-Id: <199606031941.MAA16954@samba.amazon.com>
To: Larry Masinter <masinter@parc.xerox.com>
Cc: sjk@amazon.com, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Larry Masinter writes:
 > We're at that stage of review where questions like this:
 > 
 > > If a client has determined that a given server speaks HTTP 1.1, should
 > > it really be considered illegal to omit the HOST header if an absolute
 > > URI is included in the request?
 > 
 > should be ignored. What you need to say if you expect action is
 > stronger than a question, you'll need to say why what the spec says
 > SHOULDN'T be accepted.
 > 

I think it should not be an error to process an absoluteURI when no
HOST header is present, because that requirement locks us into the
HOST header more than it has to.  Do we have the future of the
protocol clearly spelled out somewhere so that we know that there
won't be a future version of HTTP in which absoluteURIs are used
everywhere?  If so, forget the question.  Otherwise, _requiring_ HOST
even if it is not necessary and unused locks us into sending HOST
until the last 1.1 servers are retired.


 > > And what happens to a 1.0 server when it gets an absoluteURI from a
 > > 1.1 client in the request?  Isn't it the case that 1.0 servers (except
 > > for proxies) expect request URIs without the scheme, host, or port?
 > 
 > It should be clear from the spec that 1.1 clients won't send
 > absoluteURIs to 1.0 servers.
 > 
 > - Larry
 > 

Right, I see now.
I guess I just misinterpreted 

	1. If Request-URI is an absoluteURI, the host is part of the
	Request-URI. Any Host header field value in the request MUST be
	ignored.

in Jim's message out of context -- thinking any old server might
get a request with an absoluteURI, which I now see won't be true in 1.1,
although that fact could be made a little more prominent... for instance
in 5.1.2, "...HTTP/1.1 clients will only generate them in requests to
proxies"  should probably be  "...HTTP/1.1 clients MUST NOT generate
them except in requests to proxies".


--Shel
Received on Monday, 3 June 1996 12:43:09 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:32:02 EDT