On Thu, Aug 1, 2019 at 10:52 PM Andres Freund <andres@anarazel.de> wrote:
>
> On 2019-08-01 22:42:23 +0200, Julien Rouhaud wrote:
> > Sure, but it requires extra wrapper functions, and the st_changecount
> > dance when writing the new value.
>
> So? You need a wrapper function anyway, there's no way we're going to
> add all those separate pg_atomic_write* calls directly.
Ok
> I also think this proposed column should probably respect
> the track_activities GUC.
Oh indeed, I'll fix that when I'll be sure of the semantics to implement.