I’m trying to use Absinthe Relay to create Relay compatible Graphql endpoint. I am having troubles to create connections with batching.

Trying to query this, which per organization the two youngest users:

query ConnectionOrganizationListPaginationQuery([...]) {
  organizations(first: 10 etc....) {
    edges {
      node {
        name
        users(first: 2, orderBy: {age: ASC}) {
          edges {
            node {
              name
              age
            }
          [...Cursors and PageInfo...] 
}

The example below works:

In schema

  connection field :users, node_type: :user do
      arg(:search_for, list_of(:user_search), description: "List of search objects")
      arg(:order_by, list_of(:user_order), description: "List of order objects")
      arg(:filter_at, list_of(:user_filter), description: "List of filter objects")
      resolve fn
        pagination_args, %{source: organization} -> 
         args = Map.merge(pagination_args, %{filter_at: [%{organization_id: ["#{organization.id}"]}]})
         UsersResolver.list(args, %{})
      end
    end

In resolver

  def list(args, _) do
     My.App.Users.Repository.list(args, %{role: :stub_admin})
    |> Connection.from_query(&Repo.all/1, args)
  end

This leads to the infamous N=1 problem. Using field :users, list_of(:user), resolve: dataloader(My.App.Users) works for batching, but doesn’t seem to be compatible with Absinthe Relay. I just can’t find the right way to do batch loading compatible with Absinthe Relay. Is it possible at all? If so, please provide a few hints or an example

ps. I rather not load the whole list of all users in memory for pagination.
pps. Yes, I have searched myself