Andrew Dunstan <andrew@dunslane.net> writes:
> Surely sending
> copy data after the end marker is sent should be an error.
I'm unconvinced. For example, this would force a client to parse the
contents of a file it's shipping over, rather than just pushing the file
verbatim and then unconditionally adding \.
The whole \. termination business is unnecessary in protocol v3, and
eventually I'd like to deprecate and then remove it. But that's a
different discussion ...
regards, tom lane