2.15 beta - onFulfillment is not a function

[kellyselden]

From my stepping through the stack traces on both 2.14 and 2.15, it seems the RSVP catch logic may have changed a little between versions.

https://github.com/kellyselden/ember-wait-for-test-helper/pull/50/files#diff-d913131b1f61f7940eaff58ff6996ec6R35
https://travis-ci.org/kellyselden/ember-wait-for-test-helper/jobs/252511064#L1969

[bekzod]

yes it has been upgraded to 3.5.0 in 2.14 and to 3.6.1 in 2.15.beta, is there any issue with it ?

[bekzod]

I think its not rsvp but is test/promise particularly this line https://github.com/emberjs/ember.js/blob/master/packages/ember-testing/lib/test/promise.js#L65
I think its need to check if function passed, not sure but error stack leads to those lines

[rwjblue]

@stefanpenner - Can you please review https://github.com/emberjs/ember.js/blob/master/packages/ember-testing/lib/test/promise.js#L12-L14? Specifically, it seems that we cannot assume onFulfillment is actually passed to our custom Promise subclasses .then. Is this a bug in RSVP somehow, or just in isolate assuming that the first arg is always present?

[bekzod]

@stefanpenner I don't think RSVP behaving incorrectly here is gist for reproducing same issue,
both native inherited and rsvp inherited Promise behaving same way rejecting when inherited then has null onFulfillment
https://gist.github.com/bekzod/68e8fc1c739745b6a5bdb12e99badfd4 (run in node 6.8.1)

[stefanpenner]

@bekzod I agree. I don't believe RSVP is the offender here, although I am surprised upgrading causes a downstream issue at all (but I dont have time to do root cause analysis on this one), the issue is in embers code, we should just fix that guy up and cut a new beta.