Fix jest test runner

[SMJ93]

Fix jest test runner on Android and iOS.

Handle no configuration flags in the Jest runner.

Should fix #362 and #363

New up to date demo project for react native which uses jest as a test runner.

To run:

cd examples/demo-react-native-jest
yarn
npm run test-android
npm run test-ios

[SMJ93]

@rotemmiz One of the builds failed because of an issue with npm, could you manually trigger another build?

network request to https://registry.npmjs.org/babel-plugin-syntax-exponentiation-operator/-/babel-plugin-syntax-exponentiation-operator-6.13.0.tgz failed, reason: getaddrinfo ENOTFOUND registry.npmjs.org registry.npmjs.org:443

[Kureev]

Can you please separate the fix itself from example so it is easier to review?

[Kureev]

Can you stick to defaults that commander use?

[SMJ93]

Because the flags are '' the defaults will be used by the commander - similar to mocha.

[Kureev]

I'm just saying that instead of using || '' everywhere in the execSync command you can specify defaults to
program.options in the beginning of the file. It supports three parameters in .option chain:

program.option('-flag', 'description', 'defaults')

so your || '' can be re-written in a more elegant way. For instance,

program.option('-l, --log-level', 'Log level ...', '')

where third '' is a default value.

Does it make sense?

[SMJ93]

@Kureev, this PR just includes the jest test runner fix.

#370 includes the new demo project.

[SMJ93]

@Kureev , I have updated the commander defaults.

[Kureev]

🎉

[Kureev]

Hi @rotemmiz! I think this one is ready to be merged

[ferrannp]

I can confirm this is needed for Jest runner to work.

[sibelius]

let's do it

[rotemmiz]

Hey @SMJ93, thank you very much for this PR!
Sorry for taking that long to respond, we've been really busy with releasing new features in our core product.

There are two issues I wanted to discuss before we merge it:

1. I am not sure I understand what is the issue this PR fixes, what happens if Jest receives empty configuration ? argparse should be able to handle empty keys or no keys, and querying it should be done carefully (if (argparse.getArgValue('key'))). It can return either null, undefined, or empty string, but they should all resolve to false.
2. Jest integration seems to be catching pretty fast, and I want to make sure it will not get broken easily. If we indeed want to fully support it, let's add another build to our build Matrix that will run the demo projects (including both Mocha and Jest test projects)

[SMJ93]

@rotemmiz no worries!

Regarding your issues:

1. Previously, passing through empty configuration to Jest resulted in strange behaviour; the build failed and verbose was always enabled.
   To fix this we added default values so it would fallback if null, undefined etc. There are probably better ways to do fix this - I will try to find some time to look into argparse, but for the time being this fixes Jest.

2. Yeah, for us (and probably many others!) it makes sense to use the same test runner for both unit and e2e tests. Yeah that sounds like a better solution! If you have time at some point maybe we could go through the build matrix?

[rotemmiz]

Sorry @SMJ93 , I won't accept a workaround solution. We either need to understand what causes Jest to act weirdly (probably they way argparse is being consumed around argparse.getArgValue('loglevel')).

[rotemmiz]

@SMJ93 , can I use the email in your github account to invite you to our slack group ?

[SMJ93]

@rotemmiz, I understand what you're saying, but at the moment the Jest test runner is broken on both iOS and Android. This would allow people to use it for the time being while we find a better fix / understand the problem better.

Yeah sure. I'll be on after work.

[sibelius]

@rotemmiz what is missing in this PR?

[rotemmiz]

Hey @sibelius , my comment above #371 (comment). This PR will probably won't make it into master.

We are actually doing a few changes internally, that will include fixes for this issue, and a few others with Jest, as preliminaries to a bigger task we plan to publish soon, parallelization support for detox tests...

[SMJ93]

@rotemmiz, is there an ETA for this fix?

[rotemmiz]

A few breaking changes in the configuration, but we believe its a sane and more usable API than what we have today.
#423

I'm closing this PR as what it tried to solve was solved in the PR 423.