Date: 1 Apr 1996 19:32:25 -0000
Message-ID: <19960401193225.5520.qmail@koobera.math.uic.edu>
From: djb@koobera.math.uic.edu (D. J. Bernstein)
To: drums@cs.utk.edu
Subject: Re: Timezones

> zone ::= 1*ALPHA / ( ("+" / "-") 4DIGIT )

The 1*ALPHA is good, better than allowing an arbitrary atom; it covers
both 822 and the extensions that people generate.

> Implementations MUST generate numeric timezones when the correct
> local timezone is known.  Non-numeric timezones are interpreted in an
> implementation dependant fashion.

This breaks compatibility. Parsers have to correctly interpret the 822
zones, particularly GMT. But I agree that, at least for the next few
years, senders MUST NOT generate anything that wasn't in 822.

> It simplifies the grammar, and accepts things like "LDL" or "UNKN"
> without condoning them.

So you prefer LOCAL, rather than omitting the zone? ... This is a tricky
situation: we're trying to assign semantics to a behavior and deprecate
it at the same time.

---Dan