Run Teardown after each test

[jvanbaarsen]

We currently run Teardown at the end of a scenario, but whenever we fail
to load in a scenario the tests are erroring and we don't perform the
cleanup

Fixes: #1815

[jvanbaarsen]

I need to test this against a PR package since the local RWT flow is broken at the moment.

[github-actions]

📦 PR Packages

Click to Show Package Download Links

https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/create-redwood-app-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-api-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-api-server-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-auth-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-cli-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-core-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-dev-server-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-eslint-config-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-eslint-plugin-redwood-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-forms-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-internal-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-prerender-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-router-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-structure-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-testing-0.26.0-cd119e1.tgz
https://rw-pr-redwoodjs-com.s3.amazonaws.com/1818/redwoodjs-web-0.26.0-cd119e1.tgz

Install this PR by running yarn rw upgrade --pr 1818:0.26.0-cd119e1

[dac09]

I don't have full context of this, but wouldn't it make more sense to use afterAll ? I imagine that you would setup a fixture for each describe block, rather than each it block

[jvanbaarsen]

@dac09 You want to reload the scenario's on a per test basis. Otherwise they start to interfere with each other.

[thedavidprice]

Thanks for this @jvanbaarsen This looks good to me and I'd like to get it into the v0.26.0 release. Will loop in @cannikin for the final ok.

but whenever we fail
to load in a scenario the tests are erroring and we don't perform the cleanup

^^ Do either of you have any suggestions about how to test for this. And/or could you confirm you've tested this locally and it's working as intended?

[cannikin]

Yeah if you can confirm this is working as planned it looks good to me @jvanbaarsen Were you ever able to test it against a PR release?

Does this take care of the case where a scenario seed fails? We've got an open issue #1804 to address that, but this change might fix it. One way to test is to add another model to the scenario that doesn't exist:

const standard = defineScenario({
  post: {
    one: {
      title: 'First post'
    }
  },
  foo: 'bar'
})

That should throw an error like can't call method create on undefined or something similar. But afterEach() should be anticipating errors and still executing, I hope?

/cc @agiannelli this PR might have coincidentally fixed your issue as well!

[jvanbaarsen]

@thedavidprice / @cannikin Please hold back a bit with merging this. I haven't been able to properly test this just yet (Should have put this back in Draft -- edit: Turns out I'm not able to return this as a draft 🙈 )

[jvanbaarsen]

Ready for review!

You can test this by running the following steps:

• Create a new project, and create a service using rw g service UserExample (We are using the default table in schema.prisma for this)

• Run rw test api, all should be green

• Update the UserExample scenario to this:

  export const standard = defineScenario({
    userExample: {
      one: { email: 'String7883988' },
      two: { email: 'String1051631' },
    },
    foo: 'bar',
  })
  

• Run tests again, you should get an error

• Run it again and you should still see the same error (If you don't get a Unique Constraint error this fix works)

[cannikin]

Sweet! So this should close both #1815 AND #1805, correct? Whether your own code fails, or the seeding of the scenario fails, the database is cleared out looks like.

[jvanbaarsen]

Yep! Sorry @agiannelli I wasn’t aware that these issues were overlapping :(

[agiannelli]

@jvanbaarsen no need to apologize! I'm still working on getting my local up and running anyway 🙈

[jvanbaarsen]

@agiannelli Let me know if you need any help with that 👍

[thedavidprice]

Hi All! I'm looking into this now as a potential patch release. brb!!

[cannikin]

[cannikin]

In this GIF metaphor the water is the extra records left around in the database.

[cannikin]

This closes #1804 not #1805!