IntroductionComparison with mobx-state-treeClass ModelsFunctional ModelsOverviewYour first functional modelFunctional model rulesCreating a model instance and using actions/viewsAutomatic model actions for property settersComputed views with optionsLife-cycle event hooksRuntime dataAccessing the runtime checking typeGetting the Typescript types for model DataFlows (async actions)Predefined fnObject and fnArrayTree-Like StructureRoot StoresSnapshotsPatchesMaps, Sets, Dates
Action Middlewares
ContextsReferencesFrozen DataRuntime Type CheckingProperty TransformsDraftsSandboxesRedux Compatibility

Functional Models


Functional models only define the behaviors (actions/views) that can be performed over data, but separating them from the data itself, thus becoming functions with the target data as parameter.

Compared with class models, this has the distinct advantage of not requiring $modelId and $modelType as properties in the snapshots, but however comes with some disadvantages as well, namely:

  • They don't have default initializers for their properties, since they use actual data objects as data.
  • They don't support lifecycle hooks directly (although onChildAttachedTo can still be used).
  • Reconciliation is somewhat worse due to the lack of a $modelId to uniquely identify the instances.
  • Tagged runtime extra data is created lazily (upon access).

That being said, they have some use cases (for example to represent a backend response that does not include $modelId/$modelType and needs to be modified locally and eventually sent back).

Your first functional model

Functional models work over data, so first we need to declare the type of the data they will work with. Data for a todo can be defined as follows:

interface Todo {
text: string
done: boolean

or if we want runtime type checking:

const todoType = types.object(() => ({
text: types.string,
done: types.boolean,

Then we can define the set of actions/views that can act over such data:

// the string identifies this model type and must be unique across your whole application
const fnTodo = fnModel<Todo>("myCoolApp/Todo") // if not using runtime type checking
// or fnModel(todoType, "myCoolApp/Todo") if using runtime type checking
// we can also define views (computeds)
// must be parameter-less!
asString() {
return `${!this.done ? "TODO" : "DONE "} ${this.text}`
// and actions
setDone(done: boolean) {
this.done = done
setText(text: string) {
this.text = text

Functional model rules

The rules that need to be followed to declare a model are:

  • Model actions need to be used in order to be able to change such data.

Of course primitives are not the only kinds of data that a model can hold. Arrays, plain objects and other objects can be used as well.

Creating a model instance and using actions/views

An instance of the todo model above can be created like this:

const myTodo = fnTodo.create({ done: true, text: "buy some milk" })

When using actions/views you need to pass the data object as first argument:

fnTodo.setDone(myTodo, false)
const str = fnTodo.asString(myTodo)

Also note that this also holds when using actions/views within other actions/views. For example:

setAll(done: boolean, text: string) {
fnTodo.setDone(this, done)
fnTodo.setText(this, text)

Automatic model actions for property setters

Since most of the times the only action we need for a property is a setter we can use setterActions to reduce boilerplace. For example, the model above could be written as:

const fnTodo = fnModel<Todo>("myCoolApp/Todo").setterActions({
setDone: "done",
setText: "text",
const myTodo = Todo.create({ done: true, text: "buy some coffee" })
fnTodo.setDone(myTodo, false)
fnTodo.setText(myTodo, "buy milk instead")

Computed views with options

Sometimes you might need to use a custom comparison function for a computed view. You can do it like this:

asStringWithOptions: {
get() {
return ...
equals: comparer.struct // or whichever you need

Life-cycle event hooks

Functional models do not support life-cycle event hooks. However, you might somewhat get a similar functionality by using onChildAttachedTo.

Runtime data

Runtime data (data that doesn't need to be snapshotable, or that needs to be tracked in any way) can be tagged (attached) to the data object in a lazy way thanks to the tag function.

const todoTag = tag((data: Todo) => {
return {
x: 10,
// functions also supported
const tagData = todoTag.for(myTodo)
tagData.x // 10
tagData.x = 20

Note that tagged data will be created lazily, this is, it will be created the first time a tag is requested for a data object, and only once.

Accessing the runtime checking type

Assuming the fnModel method was called with a type that describes the object, it is possible to get it back.

const todoType = fnTodo.type

Getting the Typescript types for model Data

  • FnModelData<typeof fnModelType>

For example FnModelData<typeof fnTodo> would return the type of the Todo interface.

Flows (async actions)

While actions define sync model actions, async model actions are possible as well with the use of flowActions:

interface Book {
title: string
price: number
const fnBookStore = fnModel<{
books: Book[]
*fetchMyBooksAsync(token: string) {
// we use `yield* _await(X)` where we would use `await X`
// note: it is `yield*`, NOT just `yield`; `_await` is a function that has to be imported
const myBooks = yield* _await(myBackendClient.getBooks(token))
this.books = myBooks
// it can be used like this
const myBookStore = BookStore.create({ books: [] })
await myBookStore.fetchMyBooksAsync("someToken")

Predefined fnObject and fnArray

In order to work over object and arrays in a functional matter without requiring declaring custom actions you can use the already predefined fnObject and fnArray.

fnObject works over any kind of objects (including model themselves!) and offers:

  • set(obj, key, value) to set a key.
  • delete(obj, key) to delete a key.
  • call(methodName, ...args) to call a method.

fnArray works over arrays and offers:

  • set(array, index, value) to set an index.
  • delete(array, index) to delete an index.
  • setLength(array, length) to set a new length.

Plus the usual array mutation methods (pop, push, etc.).