Re: role self-revocation

Greetings,

* Tom Lane (tgl(at)sss(dot)pgh(dot)pa(dot)us) wrote:
> Stephen Frost <sfrost(at)snowman(dot)net> writes:
> > * Tom Lane (tgl(at)sss(dot)pgh(dot)pa(dot)us) wrote:
> >> Agreed, this is not something to move on quickly. We might want
> >> to think about adjusting pg_dump to use explicit GRANTED BY
> >> options in GRANT/REVOKE a release or two before making incompatible
> >> changes.
>
> > I'm with Robert on this though- folks should know already that they need
> > to use the pg_dump of the version of PG that they want to move to and
> > not try to re-use older pg_dump output with newer versions, for a number
> > of reasons and this is just another.
>
> Yeah, in an ideal world you'd do that, but our users don't always have
> the luxury of living in an ideal world. Sometimes all you've got is
> an old pg_dump file. Perhaps this behavior wouldn't mess things up
> enough to make the restored database unusable, but we need to think
> about (and test) that case while we're considering changes.

I agree it's something to consider and deal with if we're able to do so
sanely, but I disagree that we should be beholden to old dump files when
considering how to move the project forward. Further, they can surely
build and install the version of PG that goes with that dump file in a
great many cases and then dump the data out using a newer version of
pg_dump. For 5 years they could do that with a completely supported
version of PG, but we've recently agreed to make an effort to do more
here by supporting the building of even older versions on modern
systems.

Thanks,

Stephen