Re: [dmarc-ietf] cleaning up the ABNF, was I-D Action: draft-ietf-dmarc-dmarcbis-07.txt

[John Levine <[e-mail]>]

It appears that Scott Kitterman  <[e-mail]> said:
>I'm not any kind of an ABNF expert, but adding the missing "u" for dmarc-psd 
>is definitely correct and I believe the *WSP for dmarc-psd and dmarc-test are 
>correct also.

My inclination is to redo the ABNF to better match what
implementations actually do -- split the record up into tag=value pairs
and then interpret each value when used. So the ABNF would be
something like this:

 dmarc-record    = dmarc-version *WSP *(";" *WSP dmarc-tag) [ ";" ]
 
 dmarc-tag       = 1*ALPHA *WSP "=" *WSP dmarc-value *WSP
 dmarc-value     = 1*(%20-3a / %3c-7e) ; space and visible except semicolon
	; in dmarc-value leading and trailing spaces are in the WSP, not the value

Then define the specific syntax for the various values, and a rule that the "p" tag
if it is present has to go first.

If people are OK with this I can make a stab at it over the weekend.

R's,
John