Re: Replication protocol doc fix

On Fri, Jun 11, 2021 at 6:12 PM Jeff Davis <pgsql(at)j-davis(dot)com> wrote:
> On Fri, 2021-06-11 at 16:57 -0400, Robert Haas wrote:
> > My impression was that CopyBoth can be initiated either way,
>
> The docs say: "In either physical replication or logical replication
> walsender mode, only the simple query protocol can be used." Is there
> some way to initiate CopyBoth outside of walsender?

Currently, no, or at least not to my knowledge. I just meant that
there seems to be nothing in the protocol specification which prevents
CopyBothResponse from being sent in response to a query sent using the
extended protocol.

> > but if
> > you use the extended query protocol, then the result is a hopeless
> > mess, because the protocol is badly designed:
> >
> https://www.postgresql.org/message-id/CA+Tgmoa4eA+cPXaiGQmEBp9XisVd3ZE9dbvnbZEvx9UcMiw2tg@mail.gmail.com
>
> It seems like you're saying that CopyIn and CopyBoth are both equally
> broken in extended query mode. Is that right?

Yeah.

> > But I think you're correct in saying that the discard-until-Sync
> > behavior only happens if the extended query protocol is used, so I
> > agree that the current text is wrong.
>
> Should we just document how CopyBoth works with the simple query
> protocol, or should we make it match the CopyIn docs?

I think it would make sense to make it match the CopyIn docs. Possibly
the CopyOut docs should be made more similar as well.

--
Robert Haas
EDB: http://www.enterprisedb.com