Thread: pg_autovacuum -> pg_class.reloptions?

Hi,

A long time ago, Tom proposed moving the pg_autovacuum settings into
reloptions.  I know it's really late in the devel cycle but I wonder if
such a move would be acceptable at this time?  I feel it would be a good
move, for example per
http://thread.gmane.org/gmane.comp.db.postgresql.general/92643/
and the third item in
http://thread.gmane.org/gmane.comp.db.postgresql.bugs/14175

I admit I haven't even tried yet, so I would need to start working on a
patch.  This would have the advantage that

1. users would not need to come up with "disable" values for variables
they are not interested in touching, and

2. we could have more control in detecting invalid values

Opinions?

-- 
Alvaro Herrera                                http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.

Alvaro Herrera <alvherre@commandprompt.com> writes:
> A long time ago, Tom proposed moving the pg_autovacuum settings into
> reloptions.  I know it's really late in the devel cycle but I wonder if
> such a move would be acceptable at this time?

I think it's too late to be considering essentially-cosmetic changes
for 8.3, especially when you've not even started on a patch.

<rant>
If we don't start saying "no" on a regular basis, we'll never get this
thing out the door.  We are already more than two months behind the
intended schedule, and I see absolutely nothing getting done on several
of the major patches.
</rant>
        regards, tom lane