On Fri, Aug 6, 2010 at 12:31 AM, Jon Erdman (aka StuckMojo) <postgre...@thewickedtribe.net> wrote: > Description: Setting NOT NULL on inherited column turns to real > column in dump > > I think the fix here is relatively simple: make NOT NULL on an inherited > column dump as an ALTER TABLE. > > If you set NOT NULL on an inherited column in a child table, then drop the > column from the parent, it's gone from both and all is well. > > However, if you dump and restore the db, then drop the parent column, the > inherited column remains in the child table. This is a result of the NOT > NULL dumping as a column create in the child, which I assume then shadows > the inherited column after restore.
That sucks. I have a feeling it's going to be hard to fix properly without this patch: https://commitfest.postgresql.org/action/patch_view?id=312 I don't think your proposed fix will work because the NOT NULL-ness could be either inherited or not inherited. The column could even be inherited from multiple parents, some of which have a NOT NULL constraint and others of which do not. Consider: create table top1 (a int not null); create table top2 (a int); create table bottom () inherits (top1, top2); alter table bottom no inherit <something>; If <something> = top1, then bottom.a should now allow nulls, but if <something> = top2, then it should still be not null. Unfortunately, we don't do enough bookkeeping right now to distinguish those cases. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise Postgres Company -- Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-bugs
