Chris Browne <cbbrowne@acm.org> writes:
> Here's a "drafty" patch that *tries* to do this using a GUC variable;
> it passes some interactive testing.
BTW, it strikes me that a GUC variable is quite the wrong way to go
about this. The right way is a table storage parameter, a la FILLFACTOR,
so that it can be set on a per-table basis. That would also give us a
chance to fix my concern about needs_toast_table: the case where we
might need a toast table that we didn't need before is where the toast
threshold is lowered via ALTER TABLE SET, and we could reasonably make
that command recheck the situation.
regards, tom lane