Re: Bug #630: date/time storage problem: timestamp parsed

Sean Chittenden <sean@chittenden.org> writes:
> Looks like it's a "bug" in mktime() on FreeBSD: it doesn't seem to do
> so well with invalid times that happen between daylight savings
> time...  or is that a postgres thing for not kicking up an error (out
> of bounds time)?  Or should 2am PST be converted to 3am?  -sc

We've seen a *lot* of problems on various platforms with mktime
misbehaving on corner cases ... although I have to admit that failures
in PST/PDT zone are a new one on me (at least from an American
perspective, that ain't exactly a corner case).  The
DetermineLocalTimeZone() routine in datetime.c is supposed to try to
defend against the more common forms of mktime brain-damage.  Perhaps
you can suggest a way of improving it to work around this FreeBSD
problem.

            regards, tom lane