Test suite for MagicMirror²

[roramirez]

@morozgrafix Seem good sense going to a refactor of the structure.

I think we can take two way acord you mentioned.

1.- Include your test for clock module with new format for name and respective directories
2.- Do it the refactor to all tests that remain.

I really like see how you resolve the multiples instances configuration in a one tests file.

[morozgrafix]

@roramirez cool thanks. I have it all ready on my fork. Will submit PR for review.

[qistoph]

Two commits I’ve worked on for the testing.

1. Check keys in the translation files. Produces errors currently, so I’ve added .skip.
   https://github.com/qistoph/MagicMirror/commit/123392c54934e49a397d586c1fb8dbcc4cc5d12b

2. To prevent loading app.js from corrupting the Mocha test environment, I suggest to execute the app.js in a virtual environment. This can also serve as an example for future test cases where code needs to be executed in the global namespace.
   https://github.com/qistoph/MagicMirror/commit/cd8bee1371ffc6cce7b7bf44f85cd03705e4c1bd

Any thoughts before I submit a PR?

[strawberry 3.141]

@qistoph i like the idea of the sandbox

is there a way to just warn instead of error for missing translation keys?

[qistoph]

@strawberry-3.141
How about this?
https://github.com/qistoph/MagicMirror/commit/406ae4e8c37cbf7e31c89f5341d7715bacbcf0d2

			try {
				expect(fileKeys).to.deep.equal(baseKeys);
			} catch(e) {
				if (e instanceof chai.AssertionError) {
					this.skip();
				} else {
					throw e;
				}
			}

[roramirez]

@qistoph Nice work. Related with your first point there a discussion about translations of JSON file.
https://github.com/MichMich/MagicMirror/pull/679

[roramirez]

Hi Guys,

Today I’ve playing with test suite I something going me crazy

I create the follow test
https://github.com/roramirez/MagicMirror/commit/88f1f6bbd4fe6d7f3b9391993d207c7b9ebfb2e9

Two check on same test case. On my enviroment ever but ever give me fail

smirror :: ~/MagicMirror ‹without-modules-test*› » ./node_modules/mocha/bin/mocha tests/e2e/without_modules.js


  Check configuration without modules
    ✓ Show the message create file config (83ms)
    1) Show the message create file config 2


  1 passing (6s)
  1 failing

  1) Check configuration without modules Show the message create file config 2:
     Error: An element could not be located on the page using the given search parameters ("#module_2_helloworld .module-content").
      at elements("#module_2_helloworld .module-content") - getText.js:18:17
      at execute(, "isLoading", , "require") - api.js:350:19

I pushed a sample for shared with you and these are passing on Travis
https://travis-ci.org/roramirez/MagicMirror/jobs/200521954

Anybody can give me some light?

[qistoph]

Weird… don’t see anything wrong with your code and it also runs fine on my machine…

Probably not really helpful, though, sorry

[roramirez]

@qistoph I’ll test in other instance.

What S.O are you using?

[qistoph]

@roramirez what do you mean by “S.O”?

[roramirez]

@qistoph Sorry, Spanish acronymous. I mean Operating System.

On the another hand. The same commit is passed in Travis on my personal forked but in the main Mich’s repository failed.

https://travis-ci.org/roramirez/MagicMirror/builds/200750569 (Pass)
https://travis-ci.org/MichMich/MagicMirror/builds/200750793 (Fail)

It’s so weird.

[qistoph]

@roramirez my tests ran on Ubuntu.

The failed build you’re linking has timeouts. That’s the issue I added to the Trello board. We could increase the timeout a bit and see if it happens again in the next days.

Ask Mich to re-run the failed build, he should be able to click that option in Travis, being the owner of the project. See if that fails again. That resolved it for me before.

[roramirez]

@qistoph it’s pushed https://github.com/MichMich/MagicMirror/pull/716

[roramirez]

Hey guys,

I pushed up a small refactoring for tests.
https://github.com/MichMich/MagicMirror/pull/728

This is a improvement for create new tests without unnecessary repeat inside of code.

Something I wasnt think now is how can be integrate the case test created by @qistoph because this is set a app.path inside of test (It’s something I have not tried…yet)

https://github.com/MichMich/MagicMirror/blob/e62df3b3b18cc98b65656194fa5baa53b9110fbc/tests/e2e/dev_console.js#L37

[roramirez]

A little update and I requests ideas and help

Today I’m playing with a test for newsfeed module. Its not functional. I’m using the same MagicMirror as backend for response a feed (in the future can be good idea a fake server for testing purposes).

The patch can you find on
https://github.com/roramirez/MagicMirror/commit/db87f9e15b79827e947316c3ffba5b4ca49854b7

The main case is the waiting the tests for module title and titles of news. There a delay around a 1 second after the “loading…” when is show in screen. Somebody have an idea to gets this test?

[roramirez]

I fixed before mentioned https://github.com/roramirez/MagicMirror/commit/2a94ee55ccb8fa06f4d07c2a67307b5ed0a659a3

:)

[roramirez]

The last build of today.

[roramirez]

I mention this here if someone can help

If use moment-timezone for a test this fail

• Fail in Travis: https://travis-ci.org/roramirez/MagicMirror/builds/220155902
• Commit example: https://github.com/roramirez/MagicMirror/commit/05d7c9556e773bf2b941035953200ff327efb476

[roramirez]

Hi Guys, are you still alive?

Sometimes the build failed, an example here
https://travis-ci.org/MichMich/MagicMirror/jobs/227577643

Someone have some idea how to fix this?

cc @qistoph @strawberry-3-141 @morozgrafix

[qistoph]

@roramirez sorry, haven’t had much time to spend on MagicMirror lately.

My guess is that some of the tests take too long. Especially the ones actually starting MM and not just load a module.

I imagine Travis runs the tests in some sort of container with multiple containers running on the same system. If there is some heavy load at the same time as the MM test is run this might slow down our test with a certain factor. Seeing the difference in the usual runtime between tests (couple of ms vs multiple seconds) makes me think some of our tests are on the heavier end. Having a multiple second test delayed by a factor, say, 2 makes it easily run 5 seconds. Hitting the 5 sec maximum (see log).

Two possibilities I see to fix this:

1. Reduce the runtime of long running tests
2. Increase the allowed runtime before timeout

Hopefully this helps.

Best regards and thanks for keeping this thread alive!