[BUGFIX beta] fix mouseEnter/Leave event delegation w/o jQuery for SVG & IE11

[simonihmig]

IE11 does not support Node#contains() on SVG elements, so in that case a custom fallback implementation is used.

Fixes #17225

[rwjblue]

Had one failure under the /tests/?jquery=1.8.3 test run...

Test failed: EventDispatcher:  delegated event listeners work for mouseEnter on SVG elements", source:  (49)
[1124/013147.014267:INFO:CONSOLE(52)] "    Failed assertion: mouseenter event was triggered, expected: 1, but was: 0
    at Object.@test delegated event listeners work for mouseEnter on SVG elements (http://localhost:13141/ember-tests.js:19171:14)
    at Object.<anonymous> (http://localhost:13141/ember-tests.js:76563:37)
    at runTest (http://localhost:13141/qunit/qunit.js:1478:30)
    at Test.run (http://localhost:13141/qunit/qunit.js:1464:6)
    at http://localhost:13141/qunit/qunit.js:1670:12
    at advance (http://localhost:13141/qunit/qunit.js:1116:26)", source:  (52)
[1124/013147.014915:INFO:CONSOLE(52)] "    Failed assertion: Died on test #2     at generateTest (http://localhost:13141/ember-tests.js:76562:15)
    at Set.forEach (<anonymous>)
    at moduleFor (http://localhost:13141/ember-tests.js:76546:16)
    at http://localhost:13141/ember-tests.js:18961:27
    at internalRequire (http://localhost:13141/ember.debug.js:62:14)
    at Object.requireModule [as require] (http://localhost:13141/ember.debug.js:99:14)
    at http://localhost:13141/tests/?jquery=1.8.3&nolint=true:242:61: Cannot read property 'target' of undefined
TypeError: Cannot read property 'target' of undefined
    at Object.@test delegated event listeners work for mouseEnter on SVG elements (http://localhost:13141/ember-tests.js:19172:49)
    at Object.<anonymous> (http://localhost:13141/ember-tests.js:76563:37)
    at runTest (http://localhost:13141/qunit/qunit.js:1478:30)
    at Test.run (http://localhost:13141/qunit/qunit.js:1464:6)
    at http://localhost:13141/qunit/qunit.js:1670:12
    at advance (http://localhost:13141/qunit/qunit.js:1116:26)", source:  (52)

[simonihmig]

Hm, ok, I'll check this. Although I did not touch the jQuery-based implementation, so the added test seems to have uncovered this!?

[rwjblue]

@simonihmig - Yeah, looks like it...

[rwjblue]

I'd vaguely prefer to always use contains(target, related) and tweak the contains util to deal with the ternary, what do you think?

[simonihmig]

Makes sense!

[simonihmig]

Done!

[rwjblue]

I know this isn't new, but I'm curious why are we checking !related here? Can we tweak this to avoid boolean coercion? What do we expect target.relatedTarget to be? I was thinking it could only be either undefined or null, if thats correct we could (possibly in a different PR?) tweak to related === undefined && related === null...

[simonihmig]

According to the specs the uninitialized value must be null, so I changed this to a strict null check.

[simonihmig]

Had one failure under the /tests/?jquery=1.8.3 test run...

Not sure why this is happening, but the root cause seems to be with jQuery. With /tests/?jquery=1.10.0 the test is passing again!

How should we deal with that? Skip the test when using this older version of jquery? Or can we update the test suite's old jQuery version, like using the latest 1.x?

[rwjblue]

How should we deal with that?

Lets change our oldest tested jQuery to 1.10.0. Mind updating for that?

[simonihmig]

Lets change our oldest tested jQuery to 1.10.0. Mind updating for that?

@rwjblue done, in this separate PR: #17268

I will rebase this and fix the linting error once the other PR has been merged.

[rwjblue]

Perfect, #17268 is landed. Ready for rebase / linting fix now.

[simonihmig]

@rwjblue done, green! :)

[rwjblue]

Thank you very much @simonihmig!