Arc
As with the backend, you can mix and match from the features you want to use. But there is a convenience wrapper that will help you configure it all in
the form of a custom component that provides the ArcContext and configures other Arc contexts in one go.
However, if you're looking to use some of the microservice capabilities, you will have to use the ArcContext to provide the name of the
currently running microservice. Internally, Arc uses this information to add the correct headers / query string parameters to distinguish
one microservice from the other in a composition with a single ingress in front of it.
To add Arc, you simply add the following to your application:
export const App = () => {
return (
<Arc>
{/* Your application */}
</Arc>
);
};
Configuration Options
The <Arc /> component provides centralized configuration for all commands and queries in your application. These settings apply cross-cuttingly to all operations, eliminating the need to configure individual commands or queries.
| Option | Type | Purpose |
|---|---|---|
| microservice | String | Name of the microservice, which will add necessary HTTP headers on Commands and Queries |
| development | Boolean | Whether or not we're running in development, defaults to false |
| origin | String | Url for where the APIs are located, defaults to empty string and makes them relative to the documents location |
| basePath | String | Base path for the application |
| apiBasePath | String | Base path prepended to all Command and Query requests |
| httpHeadersCallback | Function | Optional callback function that returns additional HTTP headers to include with all commands, queries, and identity requests (e.g., for including cookies or authentication tokens) |
| queryTransportMethod | QueryTransportMethod |
Transport used by observable queries (ServerSentEvents or WebSocket) |
| queryDirectMode | Boolean | Whether observable queries bypass the central hub and connect directly per-query |
| queryConnectionCount | Number | Number of observable query hub connection slots |
| observableQueryTransferMode | ObservableQueryTransferMode |
Transfer mode used by useChangeStream |
| queryCacheRetentionMs | Number | How long (ms) to keep query data in cache after the last subscriber unmounts — defaults to 30000 |
Example:
export const App = () => {
return (
<Arc apiBasePath="/some/location">
{/* Your application */}
</Arc>
);
};
Microservice Support
In microservice architectures, multiple services often share a single ingress point (e.g., an API gateway or reverse proxy). The microservice property enables the Arc to route requests to the correct backend service automatically.
How It Works
When you specify a microservice name, the Arc adds this information to all HTTP requests (via headers or query parameters), allowing the ingress to route requests appropriately:
export const App = () => {
return (
<Arc microservice="user-service">
{/* Your application */}
</Arc>
);
};
All commands and queries within this application will automatically include the microservice identifier, ensuring they reach the correct backend service.
Development Configuration
For local development, you can use environment variables to dynamically configure the microservice and API paths:
export const App = () => {
const microserviceName = import.meta.env.VITE_MICROSERVICE_NAME || 'user-service';
const apiBasePath = import.meta.env.VITE_API_BASE_PATH || '/api';
const isDevelopment = import.meta.env.DEV;
return (
<Arc
microservice={microserviceName}
apiBasePath={apiBasePath}
development={isDevelopment}>
{/* Your application */}
</Arc>
);
};
.env.development example:
VITE_MICROSERVICE_NAME=user-service
VITE_API_BASE_PATH=/api
.env.production example:
VITE_MICROSERVICE_NAME=user-service
VITE_API_BASE_PATH=/api/v1
This approach allows you to:
- Run different microservices locally with different configurations
- Switch between local development and remote APIs
- Configure different API paths for different environments
- Test microservice routing without modifying code
Multiple Microservices in One Frontend
If your frontend application needs to communicate with multiple microservices, you can override the microservice configuration for specific sections of your application using nested <Arc /> components:
export const App = () => {
return (
<Arc microservice="main-service" apiBasePath="/api">
<MainContent />
{/* This section communicates with a different microservice */}
<Arc microservice="reporting-service">
<ReportingDashboard />
</Arc>
</Arc>
);
};
Query Configuration
Query-related Arc configuration (transport, direct mode, connection count, transfer mode, and headers) is documented in Queries Configuration.
HTTP Headers Callback
The httpHeadersCallback property allows you to provide additional HTTP headers that will be automatically included with all HTTP requests made by commands, queries, and identity operations. This is particularly useful for including authentication cookies, authorization tokens, or other dynamic headers.
export const App = () => {
const getHeaders = () => {
return {
'X-Custom-Header': 'custom-value',
'Authorization': `Bearer ${getAuthToken()}`,
// Include cookies or other dynamic headers
};
};
return (
<Arc
apiBasePath="/api"
httpHeadersCallback={getHeaders}>
{/* Your application */}
</Arc>
);
};
The callback function should return a HeadersInit object (compatible with the Fetch API headers) that contains the additional headers to include with each request.
Reconnecting Queries
When authentication state changes at runtime — for example when a user logs in or logs out — the existing WebSocket or Server-Sent Events connections may no longer carry the correct credentials. The reconnectQueries() method on the Arc context tears down all active observable query subscriptions, disposes the shared transport connections, and signals every useObservableQuery hook to re-subscribe through fresh connections that pick up the current cookies and headers.
Accessing reconnectQueries
Use useContext(ArcContext) inside any component rendered within <Arc>:
import { useContext } from 'react';
import { ArcContext } from '@cratis/arc.react';
export const LoginButton = () => {
const arc = useContext(ArcContext);
const handleLogin = () => {
// Set authentication cookie or token first
document.cookie = 'auth-token=abc123; path=/; SameSite=Lax';
// Then reconnect queries so new connections carry the credential
arc.reconnectQueries?.();
};
return <button onClick={handleLogin}>Log in</button>;
};
Login and logout flow
A typical authentication transition follows this pattern:
- Login: Set the authentication cookie, then call
reconnectQueries(). New transport connections are established with the cookie attached, and the identity provider picks up the authenticated user on refresh. - Logout: Call
identity.clearIdentity()to reset the client-side identity state and remove the identity cookie, then callreconnectQueries()so queries reconnect as anonymous.
import { useContext } from 'react';
import { ArcContext } from '@cratis/arc.react';
import { useIdentity } from '@cratis/arc.react/identity';
export const AuthControls = () => {
const arc = useContext(ArcContext);
const identity = useIdentity();
const login = () => {
// Set your auth cookie
document.cookie = 'auth=...; path=/; SameSite=Lax';
arc.reconnectQueries?.();
};
const logout = () => {
// Clear cookie
document.cookie = 'auth=; path=/; expires=Thu, 01 Jan 1970 00:00:00 GMT';
// Clear client-side identity
identity.clearIdentity();
// Reconnect queries as anonymous
arc.reconnectQueries?.();
};
return (
<div>
{identity.isSet
? <button onClick={logout}>Log out</button>
: <button onClick={login}>Log in</button>}
</div>
);
};
What happens internally
Calling reconnectQueries() performs three steps in order:
- Tear down subscriptions — Every cached query entry has its subscription callback torn down without evicting the cache entry. This allows hooks to detect the unsubscribed state and re-subscribe.
- Reset the shared multiplexer — The module-level multiplexer singleton (WebSocket or SSE) is disposed, so the next subscription creates a fresh transport connection.
- Bump the query version — An internal version counter increments, which is included in the dependency array of every
useObservableQueryeffect. React re-runs the effects, each hook re-subscribes, and new connections are established with the current credentials.