Use jquery loader for tests WIP

[mishal]

This is WIP for testing against more versions of JQuery.

I'm not familiar with gulp pipe system so the gulpfile is untouched.

The idea is to pass ?jquery=3.1.1 to the unit test html when running qunit. The jquery_loader.js will than simply load corresponding version from code.jquery.com.

Can something like this be accomplished with grunt pipe system?

var jQueryVersions = ['1.10.2', '1.11.1', '2.1.3', '3.1.1'];
var getUnitTestUrls = function (baseFiles, jQueryVersions) {
  var urls = [];
  for (var i = 0, count = baseFiles.length; i < count; i++) {
    var file = baseFiles[i];
    for (var v = 0; v < jQueryVersions.length; v++) {
      urls.push(file + '?jquery=' + jQueryVersions[v]);
    }
  }
  return urls;
};

getUnitTestUrls(testGlob, jQueryVersions).pipe(qunit) ?

[centi]

Well, the problem is, that using gulp.src Gulp serves local files and does not allow query strings in the path. There is a gulp-remote-src plugin, but we would have to add some static http server to serve the test files remotely.

[centi]

I'm not sure I want to implement this. Maybe if you could go all the way and figure out how to do this in Gulp I will merge it.

But I think this should be done by some sort of an integration test. And my goal is to remove the jQuery dependency completely some day anyway.

[mishal]

The tests are now using hardcoded version of jQuery (1.10.2) which was released in 2013 http://blog.jquery.com/2013/07/03/jquery-1-10-2-and-2-0-3-released/.

Using multiple jQuery versions in the tests would check that nothing is broken when using TDI with newer versions of JQuery.

Our JS tests include Grunt for which we have this already implemented.

I will try to make this work with Gulp.

Leaving jQuery as a dep is a long term goal, so for now this PR does make sense, doesn't it?

[centi]

The hardcoded version (the 1.* family) is also the most used jQuery version overall (not only in our projects). I think that this coverage in "unit" tests is good enough for now. That is the reason why this issue has not a big priority for me now.

But don't get me wrong, I still think this is a good idea and should be done sometimes eventually.

[mishal]

I give up. Gulp is really not my coffee. I tried to make it work, but with no success.

[centi]

I would leave this for now and get back to this when rethinking the tests (see #27).

Thanks for the effort anyway.