Yes, it looks like the old ordering bug problem with pg_dump, grrr. Simon, did you try this with PG 7.2? It might do a better job though if you're using PG 7.2 obviously it didn't do a better job ...
We'll be able to make this a bit easier on folks when we abandon PG 7.1 because we can then rewrite all of postgresql.sql to use "create or replace function" and then arbitrarily tell people to run it before restoring backups without worrying about errors later.
That is ... assuming PG 7.2's dump/restore is smart enough to use "create or replace". Hmmm...come to think of it, it's probably not. Drat.
