DIfferent :base_url for different sessions?

[peaceful-james]

Is it possible to use different :base_url for different sessions?

[mhanberg]

Currently, no, but that could be a new feature.

What is your use case?

[peaceful-james]

Currently, no, but that could be a new feature.

What is your use case?

My use case is a monorepo with two projects - an Elixir/Phoenix API and LiveView site running on localhost:4000 and a React SPA running on localhost:3000. I want to test full "user flows" where customers use the SPA on 3000 and staff/admin use the LiveView site on 4000. These two apps use Graphql Subscriptions and LiveView so an action by a customer may make staff view update (with no interaction).

Basically, i want to test full user stories spanning multiple realtime apps.

[mhanberg]

Gotcha, that makes sense.

Would something like this work for you?

defmodule MyAppWeb.TwoAppFeatureTest do
  use ExUnit.Case
  use Wallaby.Feature
  @spa_session [base_url: "http://localhost:3000"]
  @admin_session [base_url: "http://localhost:4000"]

  @sessions [@spa_session, @admin_session]
  feature "stuff happens", %{sessions: [spa, admin]} do
    spa |> login_user
    admin |> login_admin

    spa
    |> visit("/some-path")
    |> click(Query.button("Do Thing"))

    admin
    |> assert_text("some indicator that the user did a thing")
  end
end

[michallepicki]

Hey @mhanberg @peaceful-james , until there's a nicer API, I know there are work-arounds. I think I had a test where I needed something like that, and I leveraged the fact that it is possible to use Browser.visit/2 with absolute paths. I had the :base_url set to an empty string, and to make it easier to visit appropriate base URL depending on context, I had helper functions. Something like spa |> visit(spa_path("/some_path")). Maybe you knew that already, but maybe this note will help someone else :)

[peaceful-james]

Thanks for the information, both of you. The application I wanted to do this for has been finished for months now and I am knee-deep in a new project.
If I ever revisit this, i will be sure to try both of your techniques (although I think i have actually tried both of them). I always found that the main problem seems to be the length/duration of Wallaby tests. Small/short tests would pass but anything with extended bouts of interaction would randomly fail.