The "create function" would've failed if plpgsql wasn't created ... I think, at least!
Actually, you can see that the PG version of the OpenACS 4 database API is creating the function with the wrong (Oracle) contents, which is why I know that the query dispatcher is hosed. Either the .xql files aren't being parsed correctly (thus my nsxml comment) or the .xql files aren't there.
Or ??? I'm out of guesses at the moment.
