Re: Floating-point timestamps versus Range Types

On Mon, Oct 18, 2010 at 1:12 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> IOW I don't think the range argument holds much water for keeping float
> timestamps alive.  The on-disk-compatibility argument does, though.

Right. I think your argument that we should "do nothing" upthread is
exactly right. Deprecating float timestamps doesn't solve any real
problem. As of today, we can assume that anyone who is still using
float timestamps is doing so because they are doing in-place upgrade
from an older version. If we do nothing, the worst thing that can
possibly happen is that MAYBE they will have some difficulties if they
use floating timestamps in combination with the range types Jeff is
proposing to implement. Or, we can remove integer date time support
and categorically prevent them from using pg_upgrade whether they care
about range types or not, and whether they actually would have
experienced problems with them or not. AFAICS, that's just being
unfriendly to no purpose.

A more interesting question is whether and how we can ease the
migration path from float timestamps to integer timestamps. Even
without range types, if someone does have a UNIQUE index on a
timestamp column, could they get an error if they dump from a
float-timestamp version of PG and restore onto an integer-timestamp
version? How would we recommend that they recover from that
situation?

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company