Hi Tom!
Tom Lane [2005-06-11 13:13 -0400]:
> Martin Pitt <martin@piware.de> writes:
> > This would require that support for both int and float timestamps is
> > present in the postmaster, but wouldn't require an immediate dump and
> > reload of all databases. Would that be possible in any way?
>=20
> Given infinite amounts of work, anything is possible. However, I'd rate
> the chances of the above actually happening as about nil.
OK, I don't have any idea how complicated this would be.
> The chances of it happening in an existing stable release branch are
> flat-out zero.
That's reasonable and shouldn't happen. I did not intend to fix that
in stable Debian/Ubuntu releases as well. :-)
> I'm not sure why you are asking for this anyway: it wouldn't do
> anything to solve the immediate problem for existing users.=20=20
I just thought about how to provide a transition path if the long-term
solution really was to get rid of integer time stamps. The proposal
would have helped to be able to convert existing clusters to float
timestamps without having to rely on old versions of the postmaster
(which is really difficult to achieve in a packaging system).
> What *would* do
> something for them is to apply the patch already committed to our CVS.
>=20
> 2005-06-04 21:48 tgl
>=20
> * src/backend/commands/: variable.c (REL7_3_STABLE), variable.c
> (REL7_4_STABLE), variable.c (REL8_0_STABLE), variable.c: Code for
> SET/SHOW TIME ZONE with a fixed-interval timezone was not prepared
> for HAVE_INT64_TIMESTAMP. Per report from Guillaume Beaudoin.
Actually fixing that bug is indeed the most preferred solution. :-)
Thanks a lot for this patch. I just got the impression that integer
timestamps were doomed...
Thanks and have a nice day,
Martin
--=20
Martin Pitt http://www.piware.de
Ubuntu Developer http://www.ubuntu.com
Debian Developer http://www.debian.org