In Ember-CLI, transforms are located in app/transforms/name.js

array.js

// app/transforms/array.js
import Ember from 'ember';
import DS from 'ember-data';

export default DS.Transform.extend({
  deserialize: function(value) {
    if (Ember.isArray(value)) {
      return Ember.A(value);
    } else {
      return Ember.A();
    }
  },
  serialize: function(value) {
    if (Ember.isArray(value)) {
      return Ember.A(value);
    } else {
      return Ember.A();
    }
  }
});

object.js

// app/transforms/object.js
import Ember from 'ember';
import DS from 'ember-data';

export default DS.Transform.extend({
  deserialize: function(value) {
    if (!Ember.$.isPlainObject(value)) {
      return {};
    } else {
      return value;
    }
  },
  serialize: function(value) {
    if (!Ember.$.isPlainObject(value)) {
      return {};
    } else {
      return value;
    }
  }
});

Couldn't these transforms be replaced with calls to DS.attr() (type string intentionally omitted)?

http://emberjs.com/api/data/classes/DS.html#method_attr

DS.attr defines an attribute on a DS.Model. By default, attributes are passed through as-is, however you can specify an optional type to have the value automatically transformed. Ember Data ships with four basic transform types: string, number, boolean and date. You can define your own transforms by subclassing DS.Transform.

the real trick of this would be accurate dirty checking

@seanpdoyle DS.attr('array') and DS.attr('object') are not supported by default in ember data only "string, number, boolean and date" so you have to define the transforms yourself.

@billpull Point is, if you don't specify a type in your call to DS.attr(), they'll be passed through as-is, and just work. I just tried this out myself and it worked fine.

Sort of late to the conversation, but I find these transforms useful if you want to use defaultValue with the attr. Otherwise you are correct, the objects just pass through when no attributes type is passed in.