Re: Content-Disposition: LWS in parameter ABNF; was: ISSUE-280: whitespace in parameter syntax?

I recall that we purposefully omitted OWS around the '=' because of this.


On 07/03/2011, at 6:53 AM, Julian Reschke wrote:

> On 06.03.2011 12:19, Julian Reschke wrote:
>> Hi,
>> 
>> see <http://trac.tools.ietf.org/wg/httpbis/trac/ticket/280>...:
>> 
>> Section 2.3 of draft-ietf-httpbis-p3-payload:
>> 
>> parameter = attribute "=" value
>> attribute = token
>> value = word
>> 
>> I believe we need OWS around "=".
>> 
>> (The grammar didn't change from 2616, but we 2616 had implied LWS...).
>> 
>> Best regards, Julian
> 
> Philip Jägenstedt points out on the HTML WG mailing list that RFC 2616 said in the prose:
> 
> "Linear white space (LWS) MUST NOT be used between the type and subtype, nor between an attribute and its value."
> 
> So what HTTPbis says in P3 is consistent with that.
> 
> That being said, RFC 5987, RFC 5988 and the Content Disposition draft disagree with this, and *do* allow linear white space (RFC 5987 puts it in explicitly, RFC 5988 inherits 2616 ABNF rules, and so does C-D).
> 
> I see two ways out:
> 
> - recognize that 5987 and 5988 are wrong, raise errata, and fix C-D before it goes to the IESG
> 
> - recognize that 5987 and 5988 reflect common usage, and potentially fix it in HTTPbis
> 
> Feedback appreciated,
> 
> Julian
> 

--
Mark Nottingham   http://www.mnot.net/

Received on Monday, 7 March 2011 04:47:44 UTC