Like Ember Data, but not as good
JavaScript Ruby
Switch branches/tags
Nothing to show
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


A complex solution to a simple problem.

Why not Ember Data

Ember Data is more than likely fine for your purposes, you probably don't want to use this.

At the time of writing Ember Data didn't match my needs. It probably will do in future, but I'm impatient.

I wanted a bit more control over the URLs used and how the data is serialized/deserialized, I needed composed objects, setting properties without marking the object as dirty etc...

This doesn't handle dirty tracking and many of the other nice things you get with Ember Data, it assumes you know what you want to save and when.

It basically just provides serializing to / from JSON with an identity map and leaves most of the rest to you.


  • Tests for stores
  • Validation
  • Non-embedded associations (foreign key support, haven't needed it yet myself)
  • Use state machines for lifecycles instead of properties

Quick Overview

App.Person = Ember.Object.extend({})

App.personSchema = EmberMapper.Schema.create({
  modelClass: App.Person
  mappings: {
    firstName: "string",
    lastName: "string"

App.peopleStore = EmberMapper.Store.create({
  schema: App.PersonSchema

people = App.peopleStore.findMany({name: "Bob"})

person = App.peopleStore.find(123)
person.set("firstName", "Terry")



Your models are 'plain old ember objects', they don't know anything about the data store and don't need to inherit / include anything.

App.Person = Ember.Object.extend()

In order for the identity map to work, it needs to set a property _im_cid on your model instances.

When saving / loading etc... it will set some flags on your model:

* isLoading
* isSaving
* isUpdating
* isCreating
* isDeleting
* isDeleted


A schema uses serializers to convert a model to / from JSON.

They are identity map aware so deserializing will update the identity map for the schema.

App.PersonSchema = EmberMapper.Schema.extend({
  modelClass: "App.Person",

  mappings: {
    firstName: EmberMapper.Schema.attr("string"),         // use the string type
    lastName: "string",                                   // or to save typing, just the name
    account:"App.AccountSchema"), // map to another schema for embedded documents
    tasks: EmberMapper.Schema.many("App.TaskSchema")      // or arrays of embedded documents

  // you can include one-way mappings from the JSON
  // which is handy for counts etc... which you don't want to save back
  fromMappings: {
    numProjects: EmberMapper.Schema.attr("number"), // this won't be serialized to JSON
    createdAt: "timestamp"

  // and to JSON in occasions where you want to send something extra
  toMappings: {
    sendInvite: EmberMapper.Schema.attr("boolean") // this won't get deserialized from JSON

Model Class

Most of the time your schema is for one specific model type.

Sometimes you might want to implement polymorphism / STI where you create different kinds of models based on the JSON which is returned.

Simply override modelClassForJSON to do your bidding.

modelClassForJSON: function(json) {
  if (json.type == "car") {
    return App.Car;
  } else {
    return App.Bus;

Custom Attribute Serializing

An attribute is simply something with a from and a to method.

  • from takes JSON and turns it into an object.
  • to takes an object and turns it into JSON.

How it does that is up to you.

Any attribute on EmberMapper.Schema.transforms is available to be used for serializing / deserializing.

EmberMapper.Schema.transforms.timestamp = {
  from: function (serialized) { return new Date(serialized * 1000); },
  to: function (deserialized) { return deserialized.getTime() / 1000; }

JSON Key Naming Conventions

The default naming convention is to camelcase your JSON keys

first_name -> firstName

You can override this on a Schema by overriding propertyToKeyName

propertyNameToKey: function(key) {
  return Ember.String.underscore(key);

If you want to change it for all Schemas, reopen EmberMapper.Schema

  propertyNameToKey: function(key) {
    return key.toUpperCase();


A Store takes a Schema and loads/saves it to the intertubes.

peopleStore = EmberMapper.Store.create({
  schema: peopleSchema

Requests return immediately and are loaded when data arrives.

This returns a Person right away with its ID set to 123, it will have isLoading set to true.


This returns a RecordArray with a content of [], again isLoading will be true until data arrives.

  named: "bob"


You should set a base URL for the store:

  url: "/api/v1/people"

This is then built on for all requests, for example updateRecord will use the url with the record ID appended.

If you want to customize a specific URL, override it and do what you like:

  deleteRecordUrl: function(record) {
    return this.get("url") + "/deletification/" + record.get("id");

The url can be a computed property, so you make it bind to something else in your app:

  currentProjectBinding: "App.current.project",

  url: function(){
    var projectId = this.getPath("");
    return "/projects/"+projectId;

Custom Requests

You're encouraged to write your own custom requests. Instead of your code being littered with:

  project_id: App.getPath("")

You can have:


Simply add your own finder which calls the built in ones:

  named: function(name) {
    return this.findMany({
      name: "bob"

Or if you need to do sideloading or any other custom stuff, feel free! The xxxRequest methods return a jQuery deferred ajax object, so you can add your own callbacks.

Just look at findMany / updateRecord / etc... to write your own:

  paginated: function(page) {
    var records = EmberMapper.RecordArray.create();

    // same as usual findMany, but with your own parameters
    var ajax = this.findManyRequest(this.findManyUrl(), records, {
      data: { page: page }

    // add your own callback and do more stuff when the request completes
      records.set("pagination", {
        perPage: data.per_page

    return records;

If you want to do something to every request then you can override ajax or one of the xxxRequest methods:

  findManyRequest: function(url, records, hash) {
    var ajax = this._super(url, records, hash);
      // do stuff on error
    return ajax;