2822 revised grammar (Was Re: Malformed header - what would you do?)

On Fri July 22 2005 19:22, Frank Ellermann wrote:

> | However, where <word> occurs in this standard, it MUST NOT
> | be longer than 255 characters.

I have no idea where you got that from; it's certainly not in 2822.
 
> Does that affect any other <atom> outside of a <word> ?  I'm
> sure that it doesn't.
[...]
> > If you read CFWS as "comments and/or folding whitespace"
> > there's no problem.
> 
> This idea won't work for word = atom / quoted-string.

Last time I checked, "word" wasn't an abbreviation. 
 
>  [formatting]
> > Use abnff or deroff if you don't like it.
> 
> No nroff on my box, sorry.

Neither abnff nor deroff require nroff.  Abnff requires a C compiler
(also gperf, flex, and bison iff you want to modify the source).
There's at least one open source version of deroff that requires
just a C compiler to build.

> > The encoded-word rules are somewhat complex
> 
> Indeed, probably unnecessary for a 2822bis.  It's clear
> why you are interested to have it in a complete parser,
> but for 2822bis we better stick to the 2822 idea, no MIME.

I guess that's a matter for community discussion and consensus.
Ignoring MIME doesn't seem rational, and there needs to be some
sort of reconciliation between MIME (currently using 822 syntax
and differentiating user-defined fields from extension fields)
and 2822 (ABNF and no mention of user-defined fields).  There are
also issues related to 2822 definitions of word and phrase vs.
MIME (2047) rules regarding encoded-words "in" a phrase, especially
as that relates to encoded words and comments, which might be "in"
a phrase (depending on where CFWS appears in ABNF).  I think the
ABNF in the suggested revised grammar represents the intent, but
it really needs to be examined carefully.