[BUGFIX] Fix flushing of pending saves, that include a deleted record

[simonihmig]

Fixes #4993

Seems the regression was introduced with #4668, by refactoring from a forEach loop to a for loop, where we cannot simply return anymore: 063b3e5#diff-fd4c34d9a34dfde73cd3413128a3c973R1866

[simonihmig]

No idea why AppVeyor has failed, I cannot see any errors in their log...

[simonihmig]

Unfortunately the latest releases do not include this. As the problem is quite obvious and the fix uncontroversial, I hope this can be merged sooner than later, so pinging a few folks here... @bmac @stefanpenner @igorT @runspired @pangratz

[runspired]

👌

[bmac]

Sorry about that @simonihmig. I've been busy and missed this pr. Ember Data 2.14.1 has been published with this fix.

[simonihmig]

@bmac no problem, thanks for the quick follow up release! 👍

[stefanpenner]

this broke the build

[stefanpenner]

integration/deletedRecord - Deleting Records: Will resolve destroy and save in same loop (

Died on test #2     at Module.callback (http://localhost:4200/assets/tests.js:10865:19)
    at Module.exports (http://localhost:4200/assets/vendor.js:123:32)
    at requireModule (http://localhost:4200/assets/vendor.js:38:18)
    at TestLoader.require (http://localhost:4200/assets/test-support.js:4675:7)
    at TestLoader.loadModules (http://localhost:4200/assets/test-support.js:4667:14)
    at Function.TestLoader.load (http://localhost:4200/assets/test-support.js:4697:22)
    at http://localhost:4200/assets/test-support.js:4580:18: Assertion Failed: normalizeResponse must return a valid JSON API document:
	* One or more of the following keys must be present: "data", "errors", "meta".@ 24 ms
Source: 	
    at assert (http://localhost:4200/assets/vendor.js:16279:13)
    at Object.assert (http://localhost:4200/assets/vendor.js:27947:34)
    at normalizeResponseHelper (http://localhost:4200/assets/vendor.js:84257:20)
    at http://localhost:4200/assets/vendor.js:88804:19
    at Backburner.run (http://localhost:4200/assets/vendor.js:11296:23)
    at Backburner.join (http://localhost:4200/assets/vendor.js:11322:23)
    at http://localhost:4200/assets/vendor.js:88800:23
    at tryCatch (http://localhost:4200/assets/vendor.js:69108:14)
    at invokeCallback (http://localhost:4200/assets/vendor.js:69123:15)
    at publish (http://localhost:4200/assets/vendor.js:69091:9)
afterEach failed on Will resolve destroy and save in same loop: Assertion Failed: You can only unload a record which is not inFlight. `<person:null>`@ 25 ms
Source: 	
    at assert (http://localhost:4200/assets/vendor.js:16279:13)
    at Object.assert (http://localhost:4200/assets/vendor.js:27947:34)
    at Object.assertAgainstUnloadRecord [as unloadRecord] (http://localhost:4200/assets/vendor.js:77332:18)
    at InternalModel.send (http://localhost:4200/assets/vendor.js:83296:30)
    at InternalModel.unloadRecord (http://localhost:4200/assets/vendor.js:83063:10)
    at InternalModelMap.clear (http://localhost:4200/assets/vendor.js:84067:15)
    at IdentityMap.clear (http://localhost:4200/assets/vendor.js:84175:16)
    at Class.unloadAll (http://localhost:4200/assets/vendor.js:87780:25)
    at Class.willDestroy (http://localhost:4200/assets/vendor.js:88701:10)
    at Class.superWrapper [as willDestroy] (http://localhost:4200/assets/vendor.js:51006:22)
Expected 1 assertions, but 3 were run@ 26 ms
Source: 	
    at Module.callback (http://localhost:4200/assets/tests.js:10865:19)
    at Module.exports (http://localhost:4200/assets/vendor.js:123:32)
    at requireModule (http://localhost:4200/assets/vendor.js:38:18)
    at TestLoader.require (http://localhost:4200/assets/test-support.js:4675:7)
    at TestLoader.loadModules (http://localhost:4200/assets/test-support.js:4667:14)
    at Function.TestLoader.load (http://localhost:4200/assets/test-support.js:4697:22)
    at http://localhost:4200/assets/test-support.js:4580:18

[stefanpenner]

fixed with -> #5030 i think

[bmac]

Sorry about that @stefanpenner. We should look into setting up @homu on this repo again, because the pr looked green when I merged it.

[stefanpenner]

@bmac ya github is misleading

[simonihmig]

@bmac @stefanpenner Sorry about that! 😩

Still not sure why tests were passing locally and on Travis when PR was created? Or did something on master change meanwhile that caused this?