Re: #1028 USEFOR 3.1.2 Date: Resolved, I think.

[Frank Ellermann <nobody@xxxxxxxxxxxxxxxxx>]

Harald Tveit Alvestrand wrote:

> The main change is deleting the ABNF (unneccesary) and doing
> the import of unknown timezones in a NOTE rather than in
> normative text.

Okay.  The first statement in the note is dubious, I'd prefer
s/any/other/ resulting in:

  this specification does not change [RFC2822], which says
  that agents MUST NOT generate <date-time> constructs which
  include other zone names defined by <obs-zone>.

You already said that GMT is deprecated and MUST be accepted.
MUST NOT generate and MUST accept is a general 2822 obs-rule.
MUST NOT generate and MAY  accept is the usefor-0x  obs-rule.

If you change the latter to 2822-style for GMT it's better if
you only say "deprecated",  Besides "other zone names" can be
read as "unknown zone names", then the next statement below
your 'Note' is clearer:

> Software that accepts dates with unknown timezones SHOULD
> treat such timezones as equivalent to "-0000"
[...]

We don't want "-0000" for GMT, after s/any/other/ the "+0000"
for GMT is more obvious.

About #1022, you proposed:

| News articles MUST conform to the syntax specified in
| Section 3 of [RFC2822].  Netnews agents MAY also accept the
| obsolete syntax specified in Section 4 of [RFC2822], and
| MUST accept the "UTC" timezone (see section 3.1.2) and the
| obs-phrase construct (see [RFC2822] section 4.1) but they
| MUST NOT generate productions of such syntax.

s/"UTC"/deprecated "GMT"/ and delete "but they MUST NOT" etc.

> Trying to revoke RFC 2822's prohibition against generating
> them is, to my mind, not something we should be doing in
> this WG.

Just don't talk about it, combining MUST accept with MUST NOT
generate is odd, I prefer "SHOULD NOT generate" in such cases.

                          Bye, Frank