Nikita Shilnikov

@waiting-for-dev so I think adding a default argument to `error?` would be enough here. I can add a separate feature request about `rule.error?`.

@solnic sounds like a nice idea. I think we could have `fun` for brevity.

I'm in favor. This is definitely the place that can benefit from `Undefined`. I wish we didn't have to break things but this is for the greater good :) Mb...

@ianks how so? Using `ON CONFLICT` requires having a unique constraint anyway. If you had it before it means you weren't able to insert conflicting records even if you were...

@ianks hah, you're right, thanks ```ruby ERROR: ON CONFLICT DO UPDATE command cannot affect row a second time HINT: Ensure that no rows proposed for insertion within the same command...

Hm, at the first glance, this doesn't look convincing to me. Can you tell why you'd expect exceptions to be caught? Are there other gems that work this way?

@cutalion this should suffice ```ruby type = SQL::Types.Definition(Values::Range).constructor do |range| case range when ::Range # new format logic for ruby ranges when Values::Range format('%s%s,%s%s', range.exclude_begin? ? :'(' : :'[', range.lower,...

> I also figured out that relation returns range fields as Values::Range instead of ::Range. I expected it would return standard ruby range or at least an object with the...

> I'd like to fix all this, but would you accept a PR which will change current behavior? > I mean relation.one return a ruby range object in a range...

@cutalion this is a common practice for DB adapters to coerce input data to expected data types. `rom-sql` usually don't do anything with your data and relies on the coercion...