At Fri, 3 Jun 2016 10:52:31 +0200, Vik Fearing <vik@2ndquadrant.fr> wrote in <5751454F.6020607@2ndquadrant.fr>
> On 01/06/16 02:49, Michael Paquier wrote:
> > On Wed, Jun 1, 2016 at 3:56 AM, David G. Johnston
> > <david.g.johnston@gmail.com> wrote:
> >> On Tue, May 31, 2016 at 2:19 PM, Peter Eisentraut
> >> <peter.eisentraut@2ndquadrant.com> wrote:
> >>>
> >>> On 5/31/16 1:47 PM, Jaime Casanova wrote:
> >>>>
> >>>> Are we going to change synchronous_standby_names? Certainly the GUC is
> >>>> not *only* a list of names anymore.
> >>>>
> >>>> synchronous_standby_config?
> >>>> synchronous_standbys (adjust to correct english if necesary)?
> >>>
> >>> If the existing values are still going to be accepted, then I would leave
> >>> it as is.
> >>
> >> +1
> >
> > +1. We've made quite a lot of deal to take an approach for the N-sync
> > that is 100% backward-compatible, it would be good to not break that
> > effort.
FWIW, +1 from me.
> We could always accept it like we do for archive/hot_standby->replica.
>
> I like synchronous_standby_config, so I vote for changing it.
synchronous_standby_names is wantedly designed so as to accept
the old format. This is of couse for backward compatibility and
not to add new GUC variable needlessly.
And, I suppose that changing the domain of a GUC and changing
(only) the name of the varialbe is a bit different things and the
latter seems to me to have somewhat larger impact for users.
regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center