Andres Freund <andres@anarazel.de> writes:
> Is the generation of the duplicated error message reliable in case of a
> pg_terminate_backend()? Kinda looks like it should be?
Yeah, I think it is. libpq wouldn't notice EOF until after it
returns the backend's FATAL message.
> I guess I'll push the fix without a test for now...
Might as well push it with. If that behavior isn't stable,
it'd be nice to know.
regards, tom lane