On 2016-04-06 10:24:52 -0400, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
> > On 2016-04-06 10:15:59 -0400, Tom Lane wrote:
> >> Well, that's something worth thinking about. I assume that
> >> pg_logical_slot_get_changes could be executed in a database different from
> >> the one where a change was originated?
>
> > You can execute it, but you'll get an error:
>
> Oh good. I was afraid we had an unrecognized can o' worms here.
As posted nearby, there's a hole in that defense; for the messages
only. Pretty easy to solve though.
Allowing logical decoding from a difference would have a lot of
problems; primarily we couldn't actually look up any catalog state. But
even leaving that aside, it'd end up being pretty hard to interpret
database contents without knowledge about encoding.
Andres