W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > January to April 1995

Re: comments in HTTP headers

From: Roy T. Fielding <fielding@avron.ICS.UCI.EDU>
Date: Mon, 17 Apr 1995 18:23:30 -0700
To: Dave Kristol <dmk@allegra.att.com>
Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Message-Id: <9504171823.aa19420@paris.ics.uci.edu>
> They're a pain.  They're also ill-specified.

Pain yes, ill-specified no.

> Section 4.2 has a definition of ctext that differs from RFC 822,
> which also allows \ escapes of ( and ).

Yep.  When I wrote the initial specification of headers and comments,
I decided that allowing \ to mean escape would break too many existing
parsers.  This is still my opinion, though if a consensus says otherwise
I will change the draft.

> Also I'm unclear on which has precedence in HTTP, a comment or a
> blank line in the header.  In other words, how do I parse this:
> 
> GET / HTTP/1.0
> Accept: text/basic (this comment
> will include a blank
> [blank line]
> line)
> Accept: text/html
> [blank line]
> 
> where "[blank line]" is what it says.

Meaning what?  Only an *empty line* ends the headers of a message,
and an empty line is not allowed in a comment.  A line containing
space or HTAB is not an empty line.

 ....Roy T. Fielding  Department of ICS, University of California, Irvine USA
                                       <fielding@ics.uci.edu>
                      <URL:http://www.ics.uci.edu/dir/grad/Software/fielding>

p.s.: Conference Hell is upon us all -- I am just now getting through the
      mail amassed during the IETF trip, and am now about to embark on a
      two-week trip through California, Oregon and Washington
      (WebWorld and ICSE-17).  Henrik is in Denmark as well, so don't be
      surprised if mail goes unanswered for a while.
Received on Monday, 17 April 1995 18:35:17 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:15 EDT