Another wonderful PR, I can't not express how much I love this kind PR! Really appreciate!

I read each commit carefully, I think they're correct and beautiful. It is just I'll be more confident if we have some tests for this logic.

Is this something you would like to try? If you don't I'll try to write some tests before merging this. Don't hesitate if you have any question.

It is just I'll be more confident if we have some tests for this logic. Is this something you would like to try?

I was trying to write some test, but stuck up with IO.realTimeInstant mocking. My only solution was to provide custom Clock for IO, but it seems to be not working :(

https://scastie.scala-lang.org/rWkmMfYsTUGmRAjYg8hksQ

We can start with testing the logic inside ref.modify. Which is if we have a Map[WorkId, Work.Move] and a time at then we return the earliest non-acquired work. So, we don't have to deal with IO or mocking.

We can start with testing the logic inside ref.modify.

Oh, now I can see that we can provide moves directly through ref, not only through making request. Thanks, I will give it another try

Only thing I don't know what is the logic when two Move's createdAt equals, or can it be at all

Forgot to mention this, We don't really need to care about that case. All we need is eventually all works will be acquirred.

@lenguyenthanh sorry, I am too busy with studying and contributing for Cats (if Cats-PR got merged, I will push another PR here, sometime), would not be able to write tests any time soon ☹️

no worries @Masynchin, good luck on your cats's pr. Would always welcome your PRs!