Re: ALTER TABLE OWNER: change indexes

On Tue, 2002-03-05 at 01:50, Bruce Momjian wrote:
> Tom Lane wrote:
> > Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> > > OK, the issue with this patch is that it fixes ownership of INDEXES.
> >
> > I thought the resubmitted patch did no such thing?

That's correct.

> > > Now, we are we going with this? Can we just remove ownership of indexes
> > > totally? And sequences?
> >
> > How did you get from indexes to sequences? The issues are completely
> > different.
>
> The poster mentioned it. What does it matter? I am asking.

I think ownership is still valid for sequences (after I asked about it,
I thought about it some more -- sequences can be used by multiple
tables, and their ownership is actually used by the system).

> > I'm in favor of considering that indexes and toast tables have no
> > separate ownership, and storing zero in pg_class.relowner for them.
> > However, I have not looked to see what this might break. It might
> > be more trouble than it's worth.
>
> Well, before we reject this patch, we should decide what we are going to
> do. Of course, indexes are still in pg_class, and putting zero in there
> for a user could be trouble. It may be easier to just apply the patch.

Well, the latest version of "the patch" doesn't do much -- it just
refactors the code involved, there is no change in functionality (I
removed the ownership-changing code on Tom's request). Please apply it
-- it's quite uncontroversial. If at some later date we decide to apply
the additional ownership changing code, that's simple to do.

Cheers,

Neil

--
Neil Conway <neilconway(at)rogers(dot)com>
PGP Key ID: DB3C29FC