On Sun, Jul 24, 2011 at 11:48 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> We could perhaps go a bit further and make pqsecure_write responsible
> for the error message in non-SSL mode too, but it looks to me like
> pqSendSome has to have a switch on the errno anyway to decide whether to
> keep trying or not, so moving that responsibility would just lead to
> duplicative coding.
>
> Any objections?
No objection.
Some users we have have been confused by this message. At first I
thought it was an SSL renegotiation problem, but now I realize that
the message is a lie, so anything can be the problem. A better
message may well decrease support burden for us, and also help us find
problems...
--
fdr