Michael Paquier <michael.paquier@gmail.com> writes:
> On Tue, Dec 30, 2014 at 4:30 AM, Fabr�zio de Royes Mello
> <fabriziomello@gmail.com> wrote:
>> Shouldn't we simply leave if recovery_min_apply_delay is lower 0, and not
>> only equal to 0?
> Trivial patch for master and REL9_4_STABLE attached as long as I don't
> forget it..
It was originally intentional that the apply delay could be negative, cf
http://www.postgresql.org/message-id/52A59D10.7020209@lab.ntt.co.jp
The argument for that was completely bogus, as noted further downthread:
http://www.postgresql.org/message-id/20131212110505.GA14510@alap2.anarazel.de
but it looks like there are still residues of it in the committed patch;
both this and the totally meaningless reference to timezone differential
in the parameter's documentation.
Of course, if recovery_min_apply_delay were a proper GUC, we'd just
configure it with a minimum value of zero and be done :-(
regards, tom lane