[Info-Ingres] alter table alter column weirdness
simonl at unwired.com.au
simonl at unwired.com.au
Sun Aug 5 18:27:26 CDT 2007
On Aug 1, 10:31 pm, Karl & Betty Schendel <schen... at kbcomputer.com>
wrote:
> At 4:49 AM -0700 8/1/07, sim... at unwired.com.au wrote:
>
> >On Jul 20, 12:16 am, martin.bo... at ctsu.ox.ac.uk wrote:
> >> create table x (a varchar(20) not null not default, b integer not null not
> >> default) with page_size=4096\g
> >> alter table x alter column a varchar(20) with null not default\g
>
> >Isn't that an erroneous statement? Not sure if not default and with
> >null have ever been able to co-exist. If R3 accepted it, that would
> >be a bug in R3.
>
> I would have to disagree. If it's NOT accepted, I would consider
> that a bug. It's perfectly reasonable to have a mandatory nullable
> column, IMO. It just means that if you want to put a null into
> the column, you have to do it explicitly.
>
> Karl
You are correct. I thought it was not allowed, not sure if it has
changed somewhere along the line, but 2.6 allows it, or at least is
supposed to.
More information about the Info-Ingres
mailing list