Integration testing

[jwasilgeo]

@tomwayson please review

• adds protractor integration testing
• adds the ability to do a new gulp test
• attempts to establish a pattern for:
  • waiting on async JSAPI map properties in the DOM
  • placing some reusable utility functions among specs in the protractor conf.js (debatable)
• so far, includes specs for set-basemap, simple-map, and web-map

resolves #79

[tomwayson]

Nice work @jwasil, this is so 💰!

I've reviewed your comments about waiting for async JSAPI map properties and putting the reusable utility functions in the global.

First, I've replaced the global helper functions w/ a helper module that is loaded by the specs. I'm just not familiar enough w/ protractor to know if there is a better way to wait for async elements to be loaded on the page than the way you are doing it. It appears to be working consistently for me, so unless you are seeing the tests fail simply due to race conditions, I'd say let's go ahead and merge this. We can always refine the way those helper methods later if needed.

Also, I configured protractor to use Jasmine 2 so we could use beforeAll() to load each page only one time per spec. I like my integration specs to run as many tests as possible before having to reload the page (if at all).

Once merged into master I'll open up individual issues for each of the remaining specs as well as one for integrating the tests into the new release process.

I'm so fired up about getting these tests into master! I think it's really going to give us confidence when merging some of those long overdue PRs.

[jwasilgeo]

@tomwayson thank you for the thorough review and for making some solid enhancements!

I really like the new helper.js for the shared spec functions, rather than hitching on to Protractor's global.

And Jasmine 2's beforeAll() is ideal; the original beforeEach() page load was getting to be unnecessary and making me feel a tiny, underlying current of . This could also open up the doors for other kinds of matchers we could use (if need be).

If we want to revisit the helper functions in the future, maybe we could use Jasmine 2's done() to help us out (which I recall you pointed out to me in a separate convo). That's assuming there is a way to make it work for us when dealing with individual async element locators, rather than whole assertions. Just a thought for later on.