Nikita Shilnikov

you should be using `required(:outer_key).hash do`

why do we need `pager` as an option in the first place? What's the problem with having it as a method instead? Using instance variables beyond constructor is not desirable...

``` gordon=# select interval '2 days 23:12'; interval ----------------- 2 days 23:12:00 (1 row) gordon=# select interval '2 days 23'; interval ----------------- 2 days 00:00:23 (1 row) gordon=# select interval...

We'll be working on this, the only reason why this isn't here yet is that PG has the `search_path` setting which allows you to lookup several schemas at once. https://www.postgresql.org/docs/9.6/static/runtime-config-client.html...

I'd go with 2.0, there is a workaround for PG and in Oracle you can create a synonym. Supporting schemas may require some changes which I'd align with automigrations.

it will require rom's identifiers to support schemas for a start, then these identifiers have to be used consistently across from and rom-sql. I tried to implement it once but...

> My only worry is that it would make identifiers more complex everywhere, whereas we only need it in rom-sql. @solnic I'm not sure if this is only applicable to...

I guess yes, raising an explicit exception is better

@solnic but we qualified schemas in `rom-sql` 2.0, I believe those `.qualified` calls aren't even needed. Specs from `rom-sql` and `rom/repo` passing

Why do you think wrapping in a subquery is ideal here? In my opinion, taking current limit into consideration makes more sense.