Spec test update for define without dependencies, correction for anon_relative

[thecodedrift]

This creates 2 new tests:

• basic_empty_deps - test define() when there is dependencies defined as undefined and when dependencies are defined as []
• config_paths_relative - test that when relative paths are used, they are relative to the Module ID and not to the module's final-resolved URL.

This changes the following Library Compliance:

• RequireJS - commented basic until next rev, which resolves basic_empty_defs
• LSJS - commented basic until next rev, which resolves basic_empty_defs

No current test suites are impacted by config_paths_relative.

Referenced Issues:
This is a rollup for #13 #14 #15 #16

[rbackhouse]

So I have been looking at the new test more closely and it seems that it will be a problem with loaders that do factory-on-require processing. The test is defining modules with the assumption that the factory call will be triggered by the define call itself. This won't work for loaders that follow the factory-on-require pattern (See https://groups.google.com/forum/?fromgroups=#!searchin/amd-implement/factory-on-define/amd-implement/ot46tE-N6L0/gUzoJhpsukIJ for more details).

lsjs actually follows factory-on-require and the fix i delivered really isn't the correct solution for it.

I think the test should look more like this where "require" is used to trigger the factory call.

go(["_reporter", "require"], function(amdJS, require) {

  function emptyDeps(then) {
    define('emptyDeps', [], function() {
      amdJS.assert(arguments.length === 0, 'basic_empty_deps: [] should be treated as no dependencies instead of the default require, exports, module');
    });
    then();
  }

  function noDeps(then) {
    define('noDeps', function(require, exports, module) {
      amdJS.assert(typeof(require) === 'function', 'basic_empty_deps: no dependencies case uses require in first slot. Is a function');
      amdJS.assert(typeof(exports) === 'object', 'basic_empty_deps: no dependencies case uses exports in second slot. Is an object.');
      amdJS.assert(typeof(module) === 'object', 'basic_empty_deps: no dependencies case uses module in third slot. Is an object.');
    });
    then();
  }

  // this nesting structure ensures that the AMD define will resolve
  // before we call the next by after the tests are ran in each use
  // case. We use named define calls to ensure there are not module
  // conflicts or mismatches that can occur using anonymous modules.
  emptyDeps(function () {
    require(['emptyDeps'], function() {
        window.setTimeout(function () {
          noDeps(function () {
            require(['noDeps'], function() {
                window.setTimeout(function () {
                  amdJS.print('DONE', 'done');
                });
            });
          });
        });
    });
  });

});

[thecodedrift]

I think that's pretty valid. If someone has factory-on-require, they would unfairly fail these tests, so we should ensure their factories are given the opportunity to run. I'll put together a PR that references this change

[thecodedrift]

Done and done. Thanks @rbackhouse for the assist!