Skip to main content
Version: Next 🚧

Imperatively modifying store data

Data in Relay stores can be imperatively modified within updater functions.

When to use updaters

Complex client updates

You might provide an updater function if the changes to local data are more complex than what can be achieved by simply writing a network response to the store and cannot be handled by the declarative mutation directives.

Client schema extensions

In addition, since the network response necessarily will not include data for fields defined in client schema extensions, you may wish to use an updater to initialize data defined in client schema extensions.

Use of other APIs

Lastly, there are things you can only achieve using updaters, such as invalidating nodes, deleting nodes, finding all connections at a given field, etc.

If multiple optimistic responses modify a given store value

If two optimistic responses affect a given value, and the first optimistic response is rolled back, the second one will remain applied.

For example, if two optimistic responses each increase a story's like count by one, and the first optimistic response is rolled back, the second optimistic response remains applied. However, it is not recalculated, and the value of the like count will remain increased by two.

When not to use updaters

To trigger other side effects

You should use the onCompleted callback to trigger other side effects. onCompleted callbacks are guaranteed to be called once, but updaters and optimistic updaters can be called repeatedly.

The various types of updater functions

The useMutation and commitMutation APIs accept configuration objects which can include optimisticUpdater and updater fields. The requestSubscription and useSubscription APIs accept configuration objects which can include updater fields.

In addition, there is another API (commitLocalUpdate) which also accepts an updater function. It will be discussed in the Other APIs for modifying local data section.

Optimistic updaters vs updaters

Mutations can have both optimistic and regular updaters. Optimistic updaters are executed when a mutation is triggered. When that mutation completes or errors, the optimistic update is rolled back.

Regular updaters are executed when a mutation completes successfully.


Let's construct an example in which an is_new_comment field (which is defined in a schema extension) is set to true on a newly created Feedback object in a mutation updater.

# Feedback.graphql
extend type Feedback {
is_new_comment: Boolean
// CreateFeedback.js
import type {Environment} from 'react-relay';
import type {
} from 'CreateFeedbackMutation.graphql';

const {commitMutation, graphql} = require('react-relay');
const {ConnectionHandler} = require('relay-runtime');

function commitCreateFeedbackMutation(
environment: Environment,
input: FeedbackCreateData,
) {
return commitMutation<FeedbackCreateData>(environment, {
mutation: graphql`
mutation CreateFeedbackMutation($input: FeedbackCreateData!) {
feedback_create(input: $input) {
feedback {
# Step 1: in the mutation response, spread an updatable fragment (defined below).
# This updatable fragment will select the fields that we want to update on this
# particular feedback object.
variables: {input},

// Step 2: define an updater
updater: (store: RecordSourceSelectorProxy, response: ?CreateCommentMutation$data) => {
// Step 3: Access and nullcheck the feedback object.
// Note that this could also have been achieved with the @required directive.
const feedbackRef = response?.feedback_create?.feedback;
if (feedbackRef == null) {

// Step 3: call store.readUpdatableFragment
const {updatableData} = store.readUpdatableFragment(
// Step 4: Pass it a fragment literal, where the fragment contains the @updatable directive.
// This fragment selects the fields that you wish to update on the feedback object.
// In step 1, we spread this fragment in the query response.
fragment CreateFeedback_updatable_feedback on Feedback @updatable {
// Step 5: Pass the fragment reference.

// Step 6: Mutate the updatableData object!
updatableData.is_new_comment = true;

module.exports = {commit: commitCreateFeedbackMutation};

Let's distill what's going on here.

  • The updater accepts two parameters: a RecordSourceSelectorProxy and an optional object that is the result of reading out the mutation response.
    • The type of this second argument is a nullable version of the $data type that is imported from the generated mutation file.
    • The second argument contains just the data selected directly by the mutation argument. In other words, it will not contain any fields selected solely by spread fragments.
  • This updater is executed after the mutation response has been written to the store.
  • In this example updater, we do three things:
    • First, we spread an updatable fragment in the mutation response.
    • Second, we read out the fields selected by this fragment by calling readUpdatableFragment. This returns an updatable proxy object.
    • Third, we update fields on this updatable proxy.
  • Once this updater completes, the updates that have been recorded are written to the store, and all affected components are re-rendered.

Example 2: Updating data in response to user interactions

Let's consider the common case of updating store data in response to a user interaction. In a click handler, let's a toggle an is_selected field. This field is defined on Users in a client schema extension.

# User.graphql
extend type User {
is_selected: Boolean
// UserSelectToggle.react.js
import type {RecordSourceSelectorProxy} from 'react-relay';
import type {UserSelectToggle_viewer$key} from 'UserSelectToggle_viewer.graphql';

const {useRelayEnvironment, commitLocalUpdate} = require('react-relay');

function UserSelectToggle({ userId, viewerRef }: {
userId: string,
viewerRef: UserSelectToggle_viewer$key,
}) {
const viewer = useFragment(graphql`
fragment UserSelectToggle_viewer on Viewer {
user(user_id: $user_id) {
`, viewerRef);

const environment = useRelayEnvironment();

return <button
onClick={() => {
(store: RecordSourceSelectorProxy) => {
const userRef = viewer.user;
if (userRef == null) {

const {updatableData} = store.readUpdatableFragment(
fragment UserSelectToggle_updatable_user on User @updatable {

updatableData.is_selected = !viewer?.user?.is_selected;
{viewer?.user?.is_selected ? 'Deselect' : 'Select'} {viewer?.user?.name}

Let's distill what's going on here.

  • In a click handler, we call commitLocalUpdate, which accepts a Relay environment and an updater function. Unlike in the previous examples, this updater does not accept a second parameter because there is no associated network payload.
  • In this updater function, we access get an updatable proxy object by calling store.readUpdatableFragment, and toggle the is_selected field.
  • Like the previous example in which we called readUpdatableFragment, this can be rewritten to use the readUpdatableQuery API.

This example can be rewritten using the environment.commitPayload API, albeit without type safety.

Alternative API: readUpdatableQuery.

In the previous examples, we used an updatable fragment to access the record whose fields we want to update. This can also be possible to do with an updatable query.

If we know the path from the root (i.e. the object whose type is Query) to the record we wish to modify, we can use the readUpdatableQuery API to achieve this.

For example, we could set the viewer's name field in response to an event as follows:

// NameUpdater.react.js
function NameUpdater({ queryRef }: {
queryRef: NameUpdater_viewer$key,
}) {
const environment = useRelayEnvironment();
const data = useFragment(
fragment NameUpdater_viewer on Viewer {
const [newName, setNewName] = useState(data?.viewer?.name);
const onSubmit = () => {
commitLocalUpdate(environment, store => {
const {updatableData} = store.readUpdatableQuery(
query NameUpdaterUpdateQuery @updatable {
viewer {
const viewer = updatableData.viewer;
if (viewer != null) { = newName;

// etc
  • This particular example can be rewritten using readUpdatableFragment. However, you may prefer readUpdatableQuery for several reasons:
    • You do not have ready access to a fragment reference, e.g. if the call to commitLocalUpdate is not obviously associated with a component.
    • You do not have ready access to a fragment where we select the parent record of the record we wish to modify (e.g. the Query in this example). Due to a known type hole in Relay, updatable fragments cannot be spread at the top level.
    • You wish to use variables in the updatable fragment. Currently, updatable fragments reuse the variables that were passed to the query. This means that you cannot, for example, have an updatable fragment with fragment-local variables and call readUpdatableFragment multiple times, each time passing different variables.

Is this page useful?

Help us make the site even better by answering a few quick questions.