Re: ABNF switch: list rules

Mark Nottingham wrote:
 
> I disagree; I've seen implementations produce empty elements,  
> and "it's ugly" isn't a good enough reason to break them.
 
> That's not to say that we shouldn't discourage it.

Julian just wrote that he will do "ABNF-ication" as first step.

After that it should be clearer that those ugly empty elements
are only ugly, the combination with more than one CRLF in LWSP
is the point where it can get out of hand wrt interoperability.

> we have an open issue about whether line folding should be
> disallowed;
>    http://tools.ietf.org/wg/httpbis/trac/ticket/77

Without any folding there can't be an <obs-FWS> multi-folding,
that would simplify the #-LWS issue.  Based on proper ABNF it
should be simple to get this right, either replace each LWSP
by [FWS] and say "MUST NOT generate <obs-FWS>", or replace it
by *WSP without folding.

If folding is still allowed (for the FWS approach) we have to
check that there are no cases where adjacent FWS is implicitly
possible (resulting in 2*FWS).

 Frank

Received on Tuesday, 27 May 2008 13:42:25 UTC