W3C home > Mailing lists > Public > www-talk@w3.org > November to December 2010

Re: Semantics of Accept header with no value

From: Julian Reschke <julian.reschke@gmx.de>
Date: Fri, 05 Nov 2010 14:17:51 +0100
Message-ID: <4CD403FF.9080509@gmx.de>
To: Daniel Armak <danarmak@gmail.com>
CC: www-talk@w3.org
Hi Daniel,

you might want to follow up on httbis mailing list; see 
<http://www.w3.org/html/wg/tracker/issues/118>.

That being said:

On 03.11.2010 15:31, Daniel Armak wrote:
> Hello,
>
> I'm writing HTTP proxy software and there's a detail of RFC 2616 that I
> don't understand. Many headers (e.g. Accept-Encoding, Accept-Charset)
> have grammar of the form 1#(...). However, there are two headers -
> Accept and TE - that have grammar of the form #(...).

For Accept-Encoding: this was a bug. See 
<http://trac.tools.ietf.org/wg/httpbis/trac/ticket/25>.

> This difference seems to mean that an Accept header with no value is
> explicitly allowed and, therefore, that it has different semantics from
> a request with no Accept header at all. What are these semantics? As far
> as I can make out, a request with Accept: present but with no value
> should just result in a 406 Not Acceptable response. (Of course, some
> servers ignore the request Accept header entirely.)

Best regards, Julian
Received on Friday, 5 November 2010 13:18:34 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 5 November 2010 13:18:42 GMT