date anomalies when converting to seconds

[Kay C Lan]

On 5/24/06, Rob Cozens <rcozens at pon.net> wrote:
>
>
> > Isn't this just dateItems without the comas.
>
> No, because dateItems works only for the 100-year period defined by the
> centuryCutoff, whereas Sarah's "Julian date" approach is
> centuryCutoff-independent.
>

Dave Cragg wrote:

I don't think this is a centuryCutoff problem but a year 2038
problem. The seconds value can't be used for dates and times beyond
3:14 am on January 18, 2038, which I guess translates into issues for
the dateItems too. See here:

  http://home.netcom.com/~rogermw/Y2038.html<http://home.netcom.com/%7Erogermw/Y2038.html>

I may be wrong, but I think the centuryCutoff only affects
conversions from 2-digit dates to 4-digit dates, e.g. from "the date"
to "dateItems".

Hmmm... interesting. In my own little world I work with lots and lots of
dates and date & times, but only ± 10 years or so but I had generally been
ignoring the centuryCutoff issue as: a) it was so far away, and b) I thought
it would be just a simple fix of setting it to a different number.

The doc statement, 'The dateItems does not change ...[is] an invariant form
that won't change' without any reference to a limitation, may have led me
down the garden path to think that it didn't suffer from a year XX limit.

Thanks for pointing this out.