[Bug] Properties of `MutableArray` "does not exist" on tracked array after importing from 'ember-source/types/preview'

[johanrd]

🐞 Describe the Bug

Importing types from ember-source/types/preview breaks types of MutableArray properties on tracked arrays.

🔬 Minimal Reproduction

Clone this repo: https://github.com/johanrd/ember-typescript-reproducible-mutable-array

😕 Actual Behavior

Open app.ts, and see typescript error 2339 on line 18: this.list.pushObject(object):

Property 'pushObject' does not exist on type '{}[]'.ts(2339)

This also happens for other properties of MutableArray, e.g. firstObject, yet all other other type imports from ember-source works well, as far as I can see.

🤔 Expected Behavior

Types of MutableArray to be imported correctly – as they are from the "old" @types/ember__array package. Test that the 'old way' runs well in the same environment by simply removing the type imports on line 8:

-import 'ember-source/types/preview';

^ no errors after removing this line

🌍 Environment

• Ember: 4.12.0 and 5.1.0-beta1

[boris-petrov]

Note that this will fail at runtime unless Array's prototype-extensions are enabled (which you shouldn't do) - so the error is correct - there is no such method on a native JS array.

[johanrd]

Hi @boris-petrov, thanks for your reply.

For me, Array's prototype-extensions seems to be enabled, which also seems to be the default in ember, according to the docs:

By default, Ember.js will extend the prototypes of native JavaScript arrays to implement the Ember.Enumerable, Ember.MutableEnumerable, Ember.MutableArray and Ember.Array interfaces. This polyfills ECMAScript 5 array methods in browsers that do not implement them, adds convenience methods and properties to built-in arrays, and makes array mutations observable.

There are no errors at runtime in the browser, only at compile time from typescript.

[boris-petrov]

Interesting! I guess the documentation hasn't been updated. As they have been disabled by default for a while now.

[johanrd]

Ok, thanks, closing in favor of #20465