Test Utils

run Helper Method

While the DeepNew provided by Joist’s test factories allows ergonomically asserting against entities without awaits, it assumes that no other code (i.e. a separate EntityManager) has mutated the entities in the underlying database.

However, often it’s desirable for your code-under-test to have a “clean slate” EntityManager that starts out completely empty, and isn’t affected by your test’s own setup code / own EntityManager, to avoid missing production bugs that only passed the tests b/c of a side effect in the test’s EntityManager.

To support this, Joist provides a run function that will, given your test’s em, create a new EntityManager and run the code-under-test against it:

import { run } from "joist-test-utils";

it("creates a book", async () => {
  const em = newEntityManager();
  // Given an author
  const a = newAuthor(em);
  // When we perform the business logic
  await run(em, (em) => performPostBook(em, { title: "t1" }));
  // Then we have a new book
  expect(a.books.get.length).toEqual(1);
  // And it has the right title
  expect(a.books.get[0].title).toEqual("t1");
});

Furthermore, after the performPostBook is executed, run will automatically refresh all entities in your test’s EntityManager, so that they see the latest values that the code-under-test’s EntityManager committed to the database.

This means we can immediately assert against a.books.get without needing to load “a 2nd Author” instance for the same row, which can be really common in tests that interact with a stateful database:

const a1 = newAuthor(em);
await performPostBook(em);
// Example of what we _don't_ need to do: reload a1
await a1_2 = em.load(Author, a1.id);
expect(a.books.get.length).toEqual(1);

run accomplishes this by calling the EntityManager.refresh method, which reloads all currently-loaded entities from the database.