From: Erland Sommarskog (sommar@algonet.se)
Date: Mon Apr 01 2002 - 13:39:36 CST
Bruce Lilly <blilly@erols.com> writes:
> "Clive D.W. Feather" wrote:
> > But it introduces the far more serious problems of English month names and
> > the disaster of alphabetic time zone "names".
>
> RFC 2822 forbids use of alphabetic zone designations for generation
> of date-time (while providing for recognition in old headers).
>
> Regarding months, the "names" (actually abbreviations in all but
> one case) have been standard and unchanged since RFC 561, and
> most are derived from Latin (and date from the introduction of
> the Julian calendar). The main point is that while 04/10
> or 04-10 may be ambiguous "10 Apr", "Apr 10", "4 Oct", and "Oct 4"
> are not. See the section on date and time in
> http://www.first.org/docs/international_comms.html.
However, month names may be localized, and then you don't know what
happens. "listopad" is the name of a month in both Polish and Croatian,
but it is not the same month. And anyway, not much software will be
able to correctly interpret all those localized abbreviations.
Yeah, localization may not be permitted, but it is going to happen anyway.
On the other hand, the ISO 8601, with or without the T, is clearly
unambiguous as long as you use four digits for the year. On top of
that it is an international standard, and not a local homebrew for
a certain community.
Then again, the fact that the RFC 2822 format is in wide use is quite
convincing argument for not using something else for what, after all,
is a fairly marginal thing like the NTTP-Posting-Date parameter.
-- Erland Sommarskog, Stockholm, sommar@algonet.se