Failing test for 12475

[ef4]

This adds a failing test for #12475. It appears that obj.set('array', [a1, a2, a3]) fails to install contentKey observers on the new array value so that the subsequent changes to foo are ignored.

[stefanpenner]

I have reduced the test cases further, and also identified some related pre-existing issues.
About to jump on a plane, plan to bisect more at another time. (Unless someone is helpful and beats me to it)

[stefanpenner]

some more research, will do more (later):

• test individual 1.13.x
• also test other the high level assertion across versions, as we may actually have more issues.

function ok(assertion, message) {
  if (assertion) { return; }
  console.error(message);
}

var isWatching = Ember.isWatching;
var get = Ember.get;
var set = Ember.set;
var a1 = { foo: true,  id: 1 };
var a2 = { foo: true,  id: 2 };
var a3 = { foo: false, id: 3 };
var a4 = { foo: false, id: 4 };
var obj = { };

Ember.defineProperty(obj, 'a', Ember.computed('array.@each.foo', function() {}));

set(obj, 'array', Ember.A([a1, a2]));
get(obj, 'a'); // kick CP;

ok( isWatching(a1, 'foo'), 'BEFORE: expected a1.foo to watched');
ok( isWatching(a2, 'foo'), 'BEFORE: expected a2.foo to watched');
ok(!isWatching(a3, 'foo'), 'BEFORE: expected a3.foo to NOT watched');
ok(!isWatching(a4, 'foo'), 'BEFORE: expected a4.foo to NOT watched');

set(obj, 'array', [a3, a4]);

ok(!isWatching(a1, 'foo'), 'AFTER: expected a1.foo to NOT watched');
ok(!isWatching(a2, 'foo'), 'AFTER: expected a2.foo to NOT watched');
ok( isWatching(a3, 'foo'), 'AFTER: expected a3.foo to watched');
ok( isWatching(a4, 'foo'), 'AFTER: expected a4.foo to watched');


/*  Failing tests
 *
 *  v1.0.0: http://emberjs.jsbin.com/juluje/edit?html,output
 *  v1.1.0: http://emberjs.jsbin.com/cofufo/edit?html,output
 *  v1.2.0: http://jsbin.com/nixiqa/edit?html,output
 *  v1.3.0: http://jsbin.com/doragu/edit?html,output
 *  v1.4.0: http://jsbin.com/sevevi/edit?html,output
 *  v1.5.0: http://jsbin.com/yaduvu/edit?html,output
 *  v1.6.0: http://jsbin.com/dezacel/edit?html,output
 *  v1.7.0: http://jsbin.com/jehohov/edit?html,output
 *  v1.8.0: http://rwjblue.jsbin.com/fovoje/edit?html,output
 *  v1.9.0: http://rwjblue.jsbin.com/leroku/edit?html,output
 *  v1.10.0: http://emberjs.jsbin.com/rwjblue/309/edit
 *  v1.11.0: http://emberjs.jsbin.com/rwjblue/373/edit?html,output
 *  v1.12.0: http://jsbin.com/nidipe/edit?html,output
 *  v1.13.0: http://jsbin.com/xovivo/edit?html,output
 *  v1.13.1 -> 1.13.13 tested in ember-cli
 *  v1.0.0 -> 2.0.0-beta.4 all share:
 *   - ok(!isWatching(a1, 'foo'), 'AFTER: expected a1.foo to NOT watched');
 *   - ok(!isWatching(a2, 'foo'), 'AFTER: expected a2.foo to NOT watched');
 *
 *  225f24e... fails the first 2
 *   - ok(!isWatching(a1, 'foo'), 'AFTER: expected a1.foo to NOT watched');
 *   - ok(!isWatching(a2, 'foo'), 'AFTER: expected a2.foo to NOT watched');
 *  03b771... fails the last 2
 *    ok( isWatching(a3, 'foo'), 'AFTER: expected a3.foo to watched');
 *    ok( isWatching(a4, 'foo'), 'AFTER: expected a4.foo to watched');
 *
 *  v2.0.0-beta.5
 *  v2.0.0: http://rwjblue.jsbin.com/zesoni/edit?html,output
 *  v2.1.0: http://jsbin.com/hiqago/edit?html,output
 *  v2.2.0:  http://jsbin.com/hiqago/edit?html,output
 *  v2.3.0: http://jsbin.com/zulacu/edit?html,js,output
*   v2.0.0 -> 2.3.0 all share:
 *    ok( isWatching(a3, 'foo'), 'AFTER: expected a3.foo to watched');
 *    ok( isWatching(a4, 'foo'), 'AFTER: expected a4.foo to watched');
 */

[stefanpenner]

@ef4 when EXTENDS_PROTOTYPE is off, this must be an Ember.A( Yes it is very bad that this fails silently, but it indicates this isn't the test failure we are looking for.

[stefanpenner]

Although this does demonstrate some issues, I don't believe it demonstrates the primary issue outline din #12475. It is clear their are several compounding issues at play, I will (as time permits) continue to expand these tests, narrowing in on a reproduction of #12475.

My time has been pretty split lately, so I haven't been able to dedicate long stretches of time which is slowing down my research. But incrementally building a test suite of assumption is allowing me to pause/resume my efforts.

I hope to pin down the various issues, from their we can work on fixing them. I hope the increased test coverage will help keep this code in-check in the future.

As I have been working through this code, I have been noticing many places were we could improve performance. So that is a positive of this whole exercise.

[rwjblue]

#12475 was fixed by #12908.

Closing this PR...