> Yup, ye olde DST-transition-makes-for-funny-day-length issue. This is
> mentioned in the docs at
> http://www.ca.postgresql.org/users-lounge/docs/7.2/postgres/regres
> s-evaluation.html#AEN18363
> although I see the troublesome tests are now in horology not timestamp.
> (Docs fixed...)
>
> > Also as a side note, I can't get make check to get past initdb if I
> > configure with --enable-multibyte on CVS tip. Is there a known problem
>
> News to me --- anyone else seeing that?
I get initdb failures all the time when building CVS. You need to gmake
clean to fix some things. Try doing a gmake clean && gmake check
Chris