Wed Oct 21 19:21:54 PDT 2009
- Previous message: [Slony1-general] PGRES_FATAL_ERROR ERROR: cannot change return type of existing function
- Next message: [Slony1-general] PGRES_FATAL_ERROR ERROR: cannot change return type of existing function
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Quoting Vick Khera <vivek at khera.org>: > On Wed, Oct 21, 2009 at 11:25 AM, Christopher Browne > <cbbrowne at ca.afilias.info> wrote: >> And I need to think about that a little further; "DROP FUNCTION IF EXISTS" >> wasn't introduced until, I believe, PostgreSQL 8.2. (It's not in 8.1.) >> >> So that strategy won't work for 1.2 :-(. > > can you wrap it inside a SAVEPOINT, and then rollback to the savepoint > if it fails? not sure how you'd do that in a simple SQL script, > though :-( I don't think we had SAVEPOINT in 7.4, which is still supported... I have a patch with a strategy for this; basically just creating a little stored procedure which simulates the aspect of "DROP FUNCTION IF EXISTS" that we need. That is, the SP looks in information_schema to see if there's a function in the cluster's schema with the given name, and drops it, if there. It's not very smart; there's a scenario where it'll blow up, but I think that scenario corresponds to someone having hacked things up dramatically enough that they can't *realistically* expect UPDATE FUNCTIONS to work out well for them. Patch is sitting in a mailbox waiting for me to try it out...
- Previous message: [Slony1-general] PGRES_FATAL_ERROR ERROR: cannot change return type of existing function
- Next message: [Slony1-general] PGRES_FATAL_ERROR ERROR: cannot change return type of existing function
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-general mailing list