"Dan Langille" <dan(at)langille(dot)org> writes:
> I forgot to ask: Does this explain why the problem occurs under 7.2.3
> but not 7.3.2?
Say again? AFAIR this issue has been with us from the beginning.
Certainly the "regression tests fail on every DST transition day"
syndrome has been around for as long as I've been using Postgres.
> Is the Right Thing To Do(tm) this:
> select current_date, current_date-1;
> instead of this:
> select current_date, (current_date - interval '1 day')::date;
The "current_date-1" locution is no doubt more efficient as well
as more likely to do what you want. Still, it'd be nice if the
other way worked.
regards, tom lane