Polymorphic Pattern via (seemingly arbitrary) Mixin w/ JSONAPI

[hhff]

Hi All!

Apologies if this has been resolved, but seeing as I'm on Ember Data 1.13.9 I thought I'd report this.

I'm working on some basic polymorphic stuff with Active Model Serializers 0.10.0.rc2, and its JSONAPI Adapter.

I'm used to the "Mixin" pattern for Polymorphism in Ember, however I've noticed that this works without any hitches:

// mixins/statisticable.js

import Ember from 'ember';
import DS from 'ember-data';

export default Ember.Mixin.create();

// models/recap.js

import DS from 'ember-data';

// ---> NOTE: Nothing is mixed in here!

export default DS.Model.extend({
  statistics: DS.hasMany('statistics')
});

// models/statistic.js

import DS from 'ember-data';

export default DS.Model({
  statisticable: DS.belongsTo('statisticable', { polymorphic: true })
});

In the case above, recap.js does not have Statisticable mixed in. It simply requires that mixin:statisticable exists on the container, and doesn't check that it's mixed into the class Polymorphic class.

Without the mixin existing, ED complains that there's no model called statisticable.

So it seems like the requirement for there to be a mixin:statisticable is kinda arbitary? Or am I missing something?

[rwjblue]

You have to have either a model or a mixin when using polymorphic types. See here for the current implementation.

[hhff]

Yup - I understand (and quite like) the "official" pattern. Just seems like a stumbling block that the statisticable model (inverse to the polymorphic declaration) doesn't actually need the Mixin applied to it - it simply needs to exist.

In the above case, does the "dummy mixin" actually do anything? Again - I may be missing something.

[igorT]

In this case the statistics relationship might not be wired up correctly, because you didn't declare it on the parent class but should otherwise work

[hhff]

Thanks dudes! Recap#statistics is declared in the models/recap.js so that relation should be fine.

Again - I'm more highlighting that it's a little confusing there's no check to ensure the Mixin is actually applied to the "polymorphic" Model (Recap in this case).

Feels arbitrary that a Mixin even has to exist here. Just wanted to let you know it may be confusing for beginners!

[wecc]

I'm going to close this as it's working as intended by not applying the mixin – it just has to exist.

In #4375 we're tracking the improvement of docs around polymorphic relationships.

[hhff]

thanks @wecc !