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: <jg@w3.org>
Date: Fri, 31 May 96 14:43:57 -0400
Message-Id: <9605311843.AA04961@zorch.w3.org>
To: Ted Hardie <hardie@merlot.arc.nasa.gov>
Cc: dmk@allegra.att.com, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
! Comments after !.
!			- Jim

From:  hardie@merlot.arc.nasa.gov (Ted Hardie)
Message-Id:  <199605311805.LAA04447@merlot.arc.nasa.gov>
Subject:  Re: Rev81: COMMENT: 5.2 The Resource Identified by a Request
To:  jg@w3.org
Date:  Fri, 31 May 1996 11:05:53 -0700 (PDT)
Cc:  dmk@allegra.att.com, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com

Jim,
	Reading section 5.2 and 5.3, I had a similar question about
the interaction of the receipt of an absoluteURI and the lack of a
HOST header.  The first bullet in 5.2 can be read to imply that an
origin server which receives a Request-URI which is an absoluteURI and
no HOST header should return the resource named in the
Request-URI. This reading is possible because the first bullet states
that any HOST header MUST be ignored when the Request-URI is an
absoluteURI.  Section 14.23 states, however, that any HTTP/1.1 request
without a HOST header must get a 400 status code in the response.
 
Either we need to put an "unless the Request-URI is an absoluteUri"
into 14.23, or we need to add something to section 5.2 which makes it
clear that there must be a host header for the origin server to
ignore, even if the Request-URI is an absoluteURI.  

				regards,
					Ted Hardie 
 
!
! The original consensus seemed to be to require the host header all
! the time (per Klensin's original recommendations), but Roy successfully 
! argued us (in the Palo Alto meeting) around to a more slightly more
! loose definition, by promising he'd personally make sure that Apache correctly
! implemented the host requirements and report errors properly
! (that host information always be present), to address
! John Klensin and my strongly held opinion that it is vital that
! any errors in the host requirements get detected and reported quickly
! to the developer community so that no buggy clients become widespread
! that do not provide host informationconform.  With Apache's large server market
! share (and prompt implementation history), I became comfortable with the 
! less restrictive requirement on host.  So given that, I'm adding the 
! "unless the Request-URI is an absoluteURI" to 14.23 rather than the 
! latter solution.
!
! If Roy doesn't do it right (or do it at all...) I'll borrow John's
! infamous ax (and then hand it back to John to use on me).
!
!				- Jim
Received on Friday, 31 May 1996 11:48:44 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:32:01 EDT