John C Klensin wrote:
We need to make a decision as to where to draw the line on this,
if at all.
I want something that I can stuff into Bill's validator resulting
in a PASS. It's not really difficult for 2821bis, all non-trivial
constructs are already in a form where that's possible (I didn't
check the results for -01 and 02 yet, but I did for -00 in 2005).
Pro: Clearly, a complete ABNF cleanup, to make the set closed
and consistent and, ideally, to eliminate and formalize as many
of the comments as possible would be a nice thing.
Con: The odds of being able to do this in a reasonable amount
of time and to do so without introducing new errors --either
within 2821bis or between 2821bis and 2822[bis?] are, based on
prior experience, not high.
NAK, it would take me at most a day, it's straight forward once
folks know what they want with special cases like TLDs, trailing
dots, IDs, or multiple paths in for-clauses.
It's also a good chance to find out what they really want, _that_
might take some time, but it's not the fault of an ABNF cleanup.
my guess is that it represents an order of magnitude change in
the amount of work we otherwise still need to get done
Adding the missing 'verb = "VERB" [etc]' to any 'VERB [etc]' with
a list command = verb / otherverb / moreverbs / [...] is really
trivial. I've done that for the -00 check, and the simultaneous
2821 vs. 2821bis-00 comparison. (I stumbled over the old files
yesterday in my "projects/spf" folder where I also have 4409bis
and "dkim" incl. Hector's drafts :-)
With that you could also create a "collected ABNF". For a long
text making it even longer if that helps to read it is IMO okay.