On Mon, 2003-06-02 at 14:00, Tom Lane wrote:
> Rod Taylor <rbt@rbt.ca> writes:
> > As you can see below, after a rename the check constraint still refers
> > to 'col' and not 'newname' as pg_constraint.consrc is not updated.
>
> The same issue has always existed with regard to pg_attrdef.adsrc.
> pg_dump ought to be using the binary column not the source column, just
> as it does for default expressions.
Figured that, and I'll make the change.
> > It seems this is a problem in 7.3 as well. I believe the solution is to
> > outright remove consrc, and enable the interface to request a text
> > version of conbin on the fly.
>
> I do not think we need to remove the column.
Is it ok that the consrc column is not synch'd with conbin? What does
it provide if it doesn't match?
At very least we should be discouraging it's use so the pgadmin,
phppgadmin, etc. folks know not to be using it.
--
Rod Taylor <rbt@rbt.ca>
PGP Key: http://www.rbt.ca/rbtpub.asc