Improve testing to give better errors about alexa-room-finder HOT 1 OPEN

I'm not at a point where I fully understand this bug yet.

Within my unit tests, here's what happens.

it('should return correct response', function (done) {
  [...]
  lambdaLocal.execute({
    event: request,
    lambdaPath: './lambda/index.js',
    profileName: 'default',
    timeoutMs: 3000,
    mute: true,
    callback(err, data) {
      if (err) {
        throw err;
      } else {
        expect(data).eql(response);
        done();
      }
    },
  });
});

lambda-local basically executes the lambda function as it usually would, then calls callback(err, data) (using the same data used for context.done()). This makes it run my checks, which work if the response is correct. The issue is when the response isn't correct. Then it just times out rather than showing how the response and the data were different.

The issue seems to be that the assert clause is throwing the error in a block of the lambda-local code (where it calls callback) rather than within the Mocha file. This means it never hits done, so times out. This still gives an error, so it tells you when stuff is broken, but really not very clearly.

What I don't understand is why this callback method works for 25/28 tests, but doesn't work for the remaining 3 tests that use the requesters. The main difference is the time taken to do the request, so it feels like it might be something to do with asynchronicity, but I can't figure it out yet. My plan is to implement the future changes to the SDK, see if that fixes it, then go from there.

from alexa-room-finder.