Sergey Prokhorov

Regarding 1st note you may look how timeouts are implemented in https://github.com/devinus/poolboy (plus https://github.com/devinus/poolboy/issues/21) and https://github.com/seth/pooler. Poolboy uses gen_server's timeout while pooler implements own mechanism. 1st one is simpler and...

Regarding 2nd note: ``` +---------+ +-------------+ +-------------+ | Caller | | epgsql_sock | | PostgreSQL | +---------+ +-------------+ +-------------+ | | | | gen_server:call with timeout | | |------------------------------------->| |...

I think notes 1 and 2 are almost about the same. If we choose a correct timeout implementation there will be no need in flushing. My note n1 wasn't truncated,...

> The gen_server basically implements a state machine, which makes the code a bit hairy. Any reason why a gen_server has been preferred to a gen_fsm? Initialy it was. See...

* I'm ok with docker if you really want to invest your time in it. There is already an issue for that #127 * It would be interesting to test...

Agreed, sounds good!

Wow, I didn't know both error and "ok" can be returned at the same time. Are you sure you are not, eg, using the same connection from multiple processes at...

Yes, it for sure makes sense. I'll look at it a bit later. Do you know if the operation then succeeds in the end? Or insert is then rolled back?...

Maybe you can provide some reproducible example / test case? Because I'm not sure I fully catch the idea.

I think I got the idea ```erlang > {ok, C} = epgsql:connect(#{host => "localhost", port => 5433, username => "postgres", password => "postgres"}). > epgsql:equery(C, "CREATE SCHEMA my_schema"). {ok,[],[]} >...