В письме от 10 апреля 2018 08:55:52 пользователь David Steele написал:
> On 1/25/18 12:27 PM, Nikolay Shaplov wrote:
> > В письме от 25 января 2018 11:29:34 пользователь Tom Lane написал:
> >> Alvaro Herrera <alvherre@alvh.no-ip.org> writes:
> >>> Tom Lane wrote:
> >>>> Well, maybe the right answer is to address that. It's clear to me
> >>>> why that would happen if we store these things as reloptions on the
> >>>> toast table, but can't they be stored on the parent table?
> >>>
> >>> Actually, Nikolay provided a possible solution: if you execute ALTER
> >>> TABLE SET (toast.foobar = xyz), and a toast table doesn't exist, create
> >>> one at that point.
> >>
> >> That adds a lot of overhead if you never actually need the toast table.
> >
> > I think this overhead case is not that terrible if it is properly warned
> > ;-)>
> >> Still, maybe it's an appropriate amount of effort compared to the size
> >> of the use-case for this.
> >
> > If you came to some final conclustion, please close the commiffest item
> > with "Return with feedback" resolution, and I write another patch...
>
> I think this patch should be marked Returned with Feedback since it
> appears there is no consensus on whether it is useful or correct, so I
> have done that.
Exactly!
But I'd like to know what kind of feedback is it :-)
What conclusion have been reached (I did not got it)
Otherwise I would not know how to rewrite this patch.
I would suggest: create a TOAST relation whenever toast.* options is set, but
give a warning it this relation will not be used for data storage (i.e. there
is no toastable columns in the table)
But I need some confirmation, in order not to write patch in vain again :-)
>
> If I got it wrong I'm happy to move it to the next CF in Waiting for
> Author state instead.
>
> Thanks,
--
Do code for fun.