On 21 October 2016 at 19:38, Vladimir Gordiychuk <folyga@gmail.com> wrote:
> Craig, Andres what do you thinks about previous message?
I haven't had a chance to look further to be honest.
Since a downstream disconnect works, though it's ugly, it's not
something I can justify spending a lot of time on, and I already did
spend a lot on it in patch review/updating/testing etc.
I don't know what Andres wants, but I think CopyFail with
ERRCODE_QUERY_CANCELED is fine.
As for plugins that collect changes in memory and only send them on
commit, I'd call them "broken". Not an interesting case IMO. Don't do
that.
-- Craig Ringer http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training & Services