Re: RFC-2445: ABNF/prose mismatch on parameters

[rweiler@xxxxxxxxxxxxx (Robert Weiler)]

I just skimmed RFC2234 (ABNF), and don't think there is any implied ordering 
in the iCal rules. The relevant section is 3.2 which describes alternatives. 
The slash character '/', does not imply any order.

The confusion may arise from section 3.5 which says a sequence group has 
contents that are STRICTLY ORDERED.  However, the examples in this section  
seem to imply that ordering is on the elements in the group, not the elements 
in a sequence included in the group. In other words,

(a *( b / c ) d)

matches ad, abd, acd, abcd, acbd, etc. which I think is what we want and 
matches the current specification. What appears to have happened is that 
RFC2234 overloads the word Sequence. Replace the phrase 'Sequence Group' with 
'Token Group' (which is how I think it was meant to be read), and the problem 
goes away.

Bob Weiler
Amplitude Software


> From: Frank_Dawson@lotus.com
> Subject: Re: RFC-2445: ABNF/prose mismatch on parameters
> To: John Stracke <francis@ecal.com>
> Cc: ietf-calendar@imc.org
> Date: Thu, 22 Jul 1999 17:33:29 GMT
> X-Priority: 3 (Normal)
> X-MIMETrack: Serialize by Router on Barium/CAM/M/Lotus(Release 5.0a |May 4, 
1999) at 07/22/99 01:33:35 PM
> MIME-Version: 1.0
> List-Archive: <http://www.imc.org/ietf-calendar/mail-archive/>
> List-Unsubscribe: <mailto:ietf-calendar-request@imc.org?body=unsubscribe>
> 
> 
> John:
> It was both through the interop testing of iCalendar/iTIP/iMIP and also an
> offline comment from one of our QE folks who read the ABNF.
> 
> BTW, I sent a note to Chris Newmann, who has previously mentored the WG on
> some of the iCalendar ABNF. I asked him if he had any suggestions.
> 
> One approach could be to institutionalize in a "special" ABNF occurrence
> indicator, such as "&" that indicates any order to a set of elements. For
> example, rather than:
> 
>      token = *(a, b, c)    ...or...  *(a / b / c)
> 
> We might specify:
> 
>      token = *(a & b & c)
> 
> With an operator that we codify within the iCalendar specification. After
> all, notations do evolve.
> 
> -- Frank
> 
> 
> 
> 
>