[prev in list] [next in list] [prev in thread] [next in thread] 

List:       openbsd-tech
Subject:    Re: user/778
From:       Paul Janzen <pjanzen () foatdi ! harvard ! edu>
Date:       1999-03-30 5:18:21
[Download RAW message or body]

> I disagree with this.
> 
> Why not make it consistent with date(1) - ie assume year is in range 1969-2068?

I thought about that, but since `at' can only schedule events in the
future, it seems more sensible in this particular case to assume a 2-digit
year means the next one about to occur.

I agree this is not the typical way 2-digit dates are handled; on the other
hand, the typical way 2-digit dates are handled leaves us with a Y2100
problem that starts taking effect in 2068, assuming that the Unix epoch
goes 64-bit before 2038.  One just has to assume that people won't
intentionally try to schedule events in the past (and we take care of
people who are one year off).  It is documented in the man page, at least.

Paul Janzen.

P.S.  As long as the Unix time is 32-bit, you'll get equivalent errors for
scheduling events in 2069 as you would if you tried to schedule in 1969,
except one probably says "garbled time" and the other something about
fixing history.  In which case changing the error from "garbled time" to
"invalid time" would fix matters.

[prev in list] [next in list] [prev in thread] [next in thread]