David G. Johnston wrote:
> On Thu, Nov 5, 2015 at 2:15 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> > I'm interested in hearing opinions from multiple people about the
> > following two questions:
> >
> > 1. Is the new behavior better than the old behavior?
> > 2. Will breaking backward compatibility make too many people unhappy?
> >
> > My guess is that the answer to the first question is "yes" and that
> > the answer to the second one is "no", but this is clearly a
> > significant incompatibility, so I'd like to hear some more opinions
> > before concluding that we definitely want to do this.
>
> For #2 I'm not that concerned about turning an error case into a non-error.
Yeah, maybe I lack imagination but I don't see how the current behavior
is actually useful. We already silently do nothing when appropriate in
many other DDL commands.
> The rationale behind #1 makes sense to me. Given all the recent work on
> "IF NOT EXISTS" we obviously think that this general behavior is desirable
> and we should correct this deviation from that norm.
Agreed.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
