On Sun, Jul 11, 2004 at 09:02:06AM -0400, Bruce Momjian wrote:
> Tom just fixed this in CVS.
This particular problem, yes, but maybe we want to detect early this
sort of problems in the future. That's why I think a dump/restore/dump
and compare both dumps could be a good test.
> Alvaro Herrera wrote:
> > On Thu, Jun 24, 2004 at 11:38:59AM -0400, Tom Lane wrote:
> > > Kris Jurka <books@ejurka.com> writes:
> > > > ERROR: invalid input syntax for type timestamp with time zone:
> > > > "0097-02-16 17:32:01 BC-08"
> > >
> > > Thanks for catching this. It never came up before because we never
> > > would try to output timezone info for a BC date...
> >
> > Maybe it's a good idea to routinely test regression dump/restore/dump
> > (and compare both dumps). This could also mean trying to keep the most
> > objects in it so features tested for regression are also tested for
> > recoverability ...
--
Alvaro Herrera (<alvherre[a]dcc.uchile.cl>)
"La soledad es compañía"
