Re: Re: [COMMITTERS] pgsql: Update autovacuum to use reloptions instead of a system catalog,

On Wed, Feb 11, 2009 at 3:40 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Alvaro Herrera <alvherre(at)commandprompt(dot)com> writes:
>> However, Takahiro-san and Euler's position is that if you do this:
>> create table foo (f1 int) with (toast.fillfactor = 70);
>> alter table foo add column f2 text;
>> Then the toast table should have the fillfactor setting.
>
> Well, that might look sensible when phrased that way. But the more
> likely scenario would be that you add column f2 six months later,
> at which point there is room for pretty serious doubt that the option
> you specified way back when would still be the optimal choice. I'm
> just fine with the concept that if ADD COLUMN causes a toast table
> to get created, that table will have default reloptions. If you want
> nondefault toast reloptions, having to specify what you want after
> the table exists (and you know what's in it) seems reasonable to me.

FWIW, I don't really buy this argument. I can't see that it's all
that implausible to think that the user might be able to prognosticate
a reasonable value for a future TOAST table. The fact that they may
end up being wrong is hardly grounds to silently ignore whatever value
they tell us they want.

On the other hand, since I've never had a reason to tune this knob
myself, for TOAST tables or otherwise, I can't say I'm feeling a
violent urge to be the one to fix it.

...Robert