Re: add column .. default - Mailing list pgsql-hackers

From Rod Taylor
Subject Re: add column .. default
Date
Msg-id 1056034416.39060.22.camel@jester
Whole thread Raw
In response to Re: add column .. default  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: add column .. default  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
On Thu, 2003-06-19 at 10:42, Tom Lane wrote:
> Rod Taylor <rbt@rbt.ca> writes:
> > On Thu, 2003-06-19 at 10:05, Alvaro Herrera wrote:
> >> Sorry, I haven't read the spec, but what happens when there is a default
> >> value already and it's not NULL?  Are tuples where column =3D default
> >> updated?  Are tuples where column IS NULL updated?
>
> > We're talking about add column.  Since it's a new column, there will
> > never be a default and all entries are NULL in PostgreSQL.
>
> That's an overly simplistic analysis, given that the column may already
> exist in child tables.  I am not sure that the behavior is fully
> consistent even now in such cases, but we at least should make certain
> it doesn't become less consistent.

Right now if the column exists in the child table, the add column is
rejected.  I assume that will remain.

It is consistent with the user providing a column that inherits a column
from a parent by the same name.

--
Rod Taylor <rbt@rbt.ca>

PGP Key: http://www.rbt.ca/rbtpub.asc

pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: add column .. default
Next
From: Tom Lane
Date:
Subject: Re: add column .. default