I'm working on a tool to simplify updating the source code of database functions. To do a "revert" my plan is to store the values of pg_proc.* before updating, and then to restore pg_proc for the given oid if a revert is necessary. This raises the question,
Is it "safe" to do, UPDATE pg_proc SET <column> = <new source code> WHERE oid = <function's oid>; instead of using the "CREATE OR REPLACE FUNCTION" command? (Using pg_dump -s + pg_restore isn't feasible since it operates on the entire schema, and not just a subset of functions, important in my case.) Thanks for any advise. -- Best regards, Joel Jacobson Glue Finance
