On 2014-09-29 13:38:37 -0400, Robert Haas wrote:
> On Mon, Sep 29, 2014 at 12:05 PM, Stephen Frost <sfrost@snowman.net> wrote:
> > Lastly, I will say that I feel it'd be good to support bi-directional
> > communication as I think it'll be needed eventually, but I'm not sure
> > that has to happen now.
>
> I agree we need bidirectional communication; I just don't agree that
> the other direction should use the libpq format. The data going from
> the worker to the process that launched it is stuff like errors and
> tuples, for which we already have a wire format. The data going in
> the other direction is going to be things like plan trees to be
> executed, for which we don't. But if we can defer the issue, so much
> the better. Things will become clearer as we get closer to being
> done.
I think that might be true for your usecase, but not for others. It's
perfectly conceivable that one might want to ship tuples to a couple
bgworkers using the COPY protocol or such.
I don't think it needs to be fully implemented, but I think we should
design it a way that it's unlikely to require larger changes to the
added code from here to add it.
Greetings,
Andres Freund
-- Andres Freund http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training &
Services