Consider using Grunt for QA/build

[webpro]

Consider using Grunt for QA/build

[ifandelse]

I'm actually curious to compare Grunt, Anvil and Gulp build setups. If I can carve out some time, I'll cut a branch and see how Gulp works...

[webpro]

You could also consider an npm run approach, which might be pretty feasible for this project.

[ifandelse]

Just a quick update. I cut a Gulp branch and have a preliminary build (no minifying, no test harness, etc.) running. I'll spend some time over the next week or two fleshing this out to see how feasible it is.

[ifandelse]

I have a Gulp based build working here (gulp branch). Let me know what you think. Next step is to investigate test & example runner options, etc.

[webpro]

Great! It's so fast. Couple of things I noted:

• return h.push(t),void 0; has become return void h.push(t); somewhere in the minified version. But it's essentially the same. Probably an updated version of uglifyjs.
• No unminified version yet.
• There's an array of licenses in package.json, so that would need a fix in the banner.

As for the tests, there are some options for a quick start (potentially they can all work on the same codebase):

• gulp-mocha
• Testem: just add <script src="/testem.js"></script> (conditionally if running testem) and you can run the tests with testem -t spec/index.html however you want. Just let it watch the right files and that should be it (I don't think Gulp adds a lot here).
• mocha-phantomjs should also be fairly easy to support/set up. Basically this all that's needed in spec/index.html:

 

if (window.mochaPhantomJS) {
    mochaPhantomJS.run();
} else {
    mocha.run();
}

[ifandelse]

I've used testem before, so I pulled it down and tried it out again. It's acting very buggy, unfortunately. It's opening multiple tabs per browser, failing tests in one, passing the tests in another - frustrating. I also have the mocha-phantomjs setup working, so we may roll with that...

[ifandelse]

I've been quite happy with Gulp, and the tests are currently setup to work in browser and "headlessly". I would eventually like to either look into testem again, or some other way to farm out the tests to run on multiple browsers/versions - but I'm closing this for now and will address that as a separate issue in the future.