enhancement: support has many limit

[ken-kost]

#2012

Not sure how correct this is regarding implementation details,.
Also I feel the test is misplaced, but it felt most convenient to make the test in that file.

Contributor checklist

• Bug fixes include regression tests
• Chores
• Documentation changes
• Features include unit/acceptance tests
• Refactoring
• Update dependencies

[zachdaniel]

We will also have to ensure that in ash_sql in join.ex when we build the related subquery that we add a limit, and test this against things like filter(related.foo == bar) that the limit is applied in that context (we may need to do some work to detect said limit there 😄)

[ken-kost]

We will also have to ensure that in ash_sql in join.ex when we build the related subquery that we add a limit, and test this against things like filter(related.foo == bar) that the limit is applied in that context (we may need to do some work to detect said limit there 😄)

Could you expand on this? What would be a failing test example? The test should be in ash I assume.
From line 370 there is:

    |> Ash.Query.unset([:sort, :distinct, :select, :limit, :offset])
    |> handle_attribute_multitenancy(tenant)
    |> hydrate_refs(context[:private][:actor])
    |> then(fn query ->
      if sort? do
        Ash.Query.sort(query, relationship.sort)
      else
        Ash.Query.unset(query, :sort)
      end
    end)

should limit check be made somewhere there?

[zachdaniel]

🚀 Thank you for your contribution! 🚀

For the testing, what I was getting at is that we should now go and confirm via tests in ash_postgres that ash_sql does the right thing with these relationships, specifically when used in filters. i.e Ash.Query.filter(query, top_2_posts.name == "foo") (just an example)

[ken-kost]

Alright, I'll try it out. 👍