Make passing JS tests a deploy blocker about rmc HOT 10 CLOSED

I can look in to this Jamie, I've already been checking out Karma anyway. I'll give Testem a look as well.

from rmc.

@phleet How do you feel about phantomjs?

from rmc.

PhantomJS is really useful, and we already have it as a dependency, but it doesn't solve the whole problem. You would still need to a build a system for reporting the tests back to the console.

It's also unfortunate that it uses JavaScriptCore (Safari's JS engine), and not V8 (Chrome's JS engine).

It's actually what I ended up using at Khan Academy for rolling my own system to solve this problem, but it can be pretty involved.

I remember that this exists though if you do want to go down that alley: http://metaskills.net/mocha-phantomjs/

from rmc.

So just to be clear, we need something that can report back to the terminal with either a 0 or 1 depending on if all the tests passed, and make deployment conditional on passing the js-tests, correct?

from rmc.

Yep. The second step is really simple given the first.

Ideally it would actually print the test results to the console as well, which should be easy if you use one of the 3 prebuilt systems described (testem, karma, and mocha-phantomjs).

from rmc.

Alright, I have something that seems to work with mocha-phantomjs, assuming I understand you correctly. Since it's already a dependency I shouldn't have to include the source for it, just for mocha-phantomjs right?

from rmc.

Hmm... we currently have no package.json for listing dependencies in npm.

Judgement call @divad12, @mduan or @jswu? Should we add a package.json then add npm install to the setup scripts, or should we check in the source for mocha-phantomjs?

from rmc.

Anything wrong with the former (package.json + npm install)?

from rmc.

it's just another thing that can get out of sync, but it shouldn't be a problem.

package.json + npm install it is.

from rmc.

I think this can be closed now, I'll reopen it if anything was missed.

from rmc.