"Marc G. Fournier" <scrappy@hub.org> writes:
> Tom, my understanding is that CONVERT stuff required an initdb, and was
> also the only thing 'critical' that went in ... correct? What impact does
> that have? For instance, could it cause one of the regression tests to
> fail, or is it something that is relatively benign?
I would like to think it's a pretty safe change, but that's why we do
betas ;-).
More seriously, we are still attacking various portability issues, and
there is still some undone docs work. I doubt we can make a tarball
tomorrow that is an honest release candidate. Maybe Monday?
regards, tom lane