On Tue, Oct 23, 2012 at 11:18:48AM -0300, Alvaro Herrera wrote:
> Please see downthread -- there is some commentary from Noah ([1] and
> others) about the patch itself. As far I understand, some changes are
> still needed, and I don't know if the last version submitted is the
> version that should be backpatched.
We'd best use the same patch for both HEAD and back branches; pg_upgrade would
carry forward faulty indexes. pg_upgrade could instead invalidate them, but I
don't think the gravity of the problem calls for it. A release note
suggesting a REINDEX of affected indexes ought to suffice.