On Feb 20, 2010, at 15:03, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Well, I'm willing to implement CREATE OR REPLACE LANGUAGE if people
> are agreed that that's a reasonable fix. I'm slightly worried about
> the restore-could-change-ownership issue, but I think that's much less
> likely to cause problems than embedding special cases for plpgsql in a
> pile of places that we'll never find again.
Just throwing this out there: would a syntax such as CREATE OF NOT
EXISTS, a complement to DROP IF EXISTS, avoid the permissions issue?
And if so, would that be a syntax we'd want to accept in general?
Could the be a CREATE IF NOT EXISTS TABLE?
Best,
David