Assists a GenServer in managing in-memory caches.

By annotating the entities to be managed, manage/5 can handle updating the cache for the given record and its associated records. (If associations are not preloaded, they will be automatically fetched.) In addition, entites with :subscribe and :unsubscribe functions defined will be automatically subscribed to and unusbscribed from as the first reference appears and the last one is dropped.

example

Example

This module owns and is responsible for affecting changes on the Car with id 1. It subscribes to updates to Person records as they may be updated elsewhere.

defmodule MyApp.CarManager do
  use GenServer
  use Indexed.Managed, repo: MyApp.Repo
  alias MyApp.{Car, Person, Repo}

  managed :cars, Car, children: [:passengers], manage_path: :passengers

  managed :people, Person,
    subscribe: &MyApp.subscribe_to_person/1,
    unsubscribe: &MyApp.unsubscribe_from_person/1

  @impl GenServer
  def init(_), do: {:ok, warm(:cars, Repo.get(Car, 1))}

  @impl GenServer
  def handle_call(:get, _from state) do
    {:reply, get(state, :cars, 1)}
  end

  def handle_call({:update, params}, _from, state) do
    case state |> get(:cars, 1) |> Car.changeset(params) |> Repo.update() do
      {:ok, new_car} = ok -> {:reply, ok, manage(state, :cars, :update, new_car)}
      {:error, _} = err -> {:reply, err, state}
    end
  end

  @impl GenServer
  def handle_info({MyApp, [:person, :update], person}, state) do
    {:noreply, manage(state, :people, :update, person)}
  end
end

managed-macro

Managed Macro

For each managed entity, the name (eg. :cars) and module (eg. MyApp.Car) must be specified. If needed, a keyword list of options should follow.

• children : Keyword list with association fields as keys and assoc_spec/0s as vals. This is used when recursing in manage/5 as well as when resolving. If an undeclared association is resolved, Repo.get/2 will be used as a fallback. Will automatically include fields given in manage_path.
• fields : List of fields which should be sortable (ascending and descending). See Indexed.Entity.
• id_key : Specifies how to find the id for a record. It can be an atom field name to access, a function, or a tuple in the form {module, function_name}. In the latter two cases, the record will be passed in. Default :id.
• manage_path : Default associations to traverse for manage/5. Also, this is the preload argument when true is used.
• prefilters - List of atoms indicating which fields should be prefiltered on. This means that separate indexes will be managed for each unique value for each of these fields, across all records of this entity type. Each two-element tuple has the field name atom and a keyword list of options.
• query_fn : Optional function which takes a queryable and returns a queryable. This allows for extra query logic to be added such as populating virtual fields. Invoked by manage/5 when the association is needed.
• :subscribe and unsubscribe : Functions which take a record's ID and manage the subscription. These must both be declared or neither.

tips

Tips

If you want to import Ecto.Query, you'll find that its preload/3 conflicts with Managed. Since Managed will use the repo as a fallback, you can exclude it this way.

defmodule MyModule do
  use Indexed.Managed
  import Ecto.Query, except: [preload: 2, preload: 3]
end