On Saturday, July 21, 2018, PG Bug reporting form <noreply(at)postgresql(dot)org>
wrote:
> I can accept either behaviour
> (though the former is more useful); it's the fact that I get both that's
> unwelcome.
>
But each of those behaviors is consistently encountered every time the
query is run so there isn't any hidden danger involved here; compared to if
the error only occurred if certain data was encountered during execution.
As Tom said, while a bit inconcsistent the effort to fix outweighs the the
fact there is no actual problem, just an unexpected dependency on the
written query.
David J.