refactor: replace welcome message with server_info status payload

Unify the initial connection handshake and capability update paths.
The server now sends a server_info status message on connect and
whenever capabilities change, enabling the onboarding flow where
voice becomes available after models are configured.
This commit is contained in:
Mohamed Boudra
2026-03-12 20:38:03 +07:00
parent 554017b0b8
commit b09a731d85
16 changed files with 177 additions and 162 deletions
@@ -29,7 +29,7 @@ function createMockTransport() {
let onOpen: () => void = () => {}
let onClose: (_event?: unknown) => void = () => {}
let onError: (_event?: unknown) => void = () => {}
let welcomeOrdinal = 1
let serverInfoOrdinal = 1
const transport: DaemonTransport = {
send: (data) => sent.push(data),
@@ -61,11 +61,16 @@ function createMockTransport() {
sent.length = 0
onMessage(
JSON.stringify({
type: 'welcome',
serverId: `srv_test_${welcomeOrdinal++}`,
hostname: null,
version: null,
resumed: false,
type: 'session',
message: {
type: 'status',
payload: {
status: 'server_info',
serverId: `srv_test_${serverInfoOrdinal++}`,
hostname: null,
version: null,
},
},
})
)
},
+24 -19
View File
@@ -4,10 +4,11 @@ import {
AgentCreatedStatusPayloadSchema,
AgentRefreshedStatusPayloadSchema,
AgentResumedStatusPayloadSchema,
parseServerInfoStatusPayload,
RestartRequestedStatusPayloadSchema,
ShutdownRequestedStatusPayloadSchema,
SessionInboundMessageSchema,
type WSWelcomeMessage,
type ServerInfoStatusPayload,
WSOutboundMessageSchema,
} from '../shared/messages.js'
import type {
@@ -450,7 +451,7 @@ export class DaemonClient {
private readonly logServerId: string | null
private readonly logClientIdHash: string
private readonly logGeneration: number | null
private lastWelcomeMessage: WSWelcomeMessage | null = null
private lastServerInfoMessage: ServerInfoStatusPayload | null = null
constructor(private config: DaemonClientConfig) {
this.logger = config.logger ?? consoleLogger
@@ -565,7 +566,7 @@ export class DaemonClient {
const transportUrl = this.resolveTransportUrlForAttempt()
const transport = transportFactory({ url: transportUrl, headers })
this.transport = transport
this.lastWelcomeMessage = null
this.lastServerInfoMessage = null
this.updateConnectionState({
status: 'connecting',
@@ -700,7 +701,7 @@ export class DaemonClient {
this.clearWaiters(new Error('Daemon client closed'))
this.rejectPendingSendQueue(new Error('Daemon client closed'))
this.terminalStreams.clearAll()
this.lastWelcomeMessage = null
this.lastServerInfoMessage = null
this.updateConnectionState(
{ status: 'disposed' },
{ event: 'DISPOSE', reason: 'Client closed', reasonCode: 'disposed' }
@@ -2796,8 +2797,8 @@ export class DaemonClient {
return requestId ?? crypto.randomUUID()
}
getLastWelcomeMessage(): WSWelcomeMessage | null {
return this.lastWelcomeMessage
getLastServerInfoMessage(): ServerInfoStatusPayload | null {
return this.lastServerInfoMessage
}
private resolveTransportUrlForAttempt(): string {
@@ -2962,18 +2963,6 @@ export class DaemonClient {
return
}
if (parsed.data.type === 'welcome') {
this.lastWelcomeMessage = parsed.data
this.resetConnectTimeout()
this.reconnectAttempt = 0
this.updateConnectionState({ status: 'connected' }, { event: 'HELLO_WELCOME' })
this.resubscribeCheckoutDiffSubscriptions()
this.resubscribeTerminalDirectorySubscriptions()
this.flushPendingSendQueue()
this.resolveConnect()
return
}
this.handleSessionMessage(parsed.data.message)
}
@@ -3050,7 +3039,7 @@ export class DaemonClient {
this.clearWaiters(new Error(reason ?? 'Connection lost'))
this.rejectPendingSendQueue(new Error(reason ?? 'Connection lost'))
this.terminalStreams.clearAll()
this.lastWelcomeMessage = null
this.lastServerInfoMessage = null
if (wasDisposed) {
this.rejectConnect(new Error(reason ?? 'Daemon client is disposed'))
@@ -3084,6 +3073,22 @@ export class DaemonClient {
}
private handleSessionMessage(msg: SessionOutboundMessage): void {
if (msg.type === 'status') {
const serverInfo = parseServerInfoStatusPayload(msg.payload)
if (serverInfo) {
this.lastServerInfoMessage = serverInfo
if (this.connectionState.status === 'connecting') {
this.resetConnectTimeout()
this.reconnectAttempt = 0
this.updateConnectionState({ status: 'connected' }, { event: 'HELLO_SERVER_INFO' })
this.resubscribeCheckoutDiffSubscriptions()
this.resubscribeTerminalDirectorySubscriptions()
this.flushPendingSendQueue()
this.resolveConnect()
}
}
}
if (msg.type === 'terminal_stream_exit') {
this.terminalStreams.clearStream({ streamId: msg.payload.streamId })
}
@@ -335,17 +335,16 @@ describe("daemon client E2E", () => {
}
}, 30000);
test("receives welcome on websocket connect", async () => {
test("receives server_info on websocket connect", async () => {
const client = new DaemonClient({
url: `ws://127.0.0.1:${ctx.daemon.port}/ws`,
clientId: `cid-e2e-${randomUUID()}`,
clientType: "cli",
});
await client.connect();
const welcome = client.getLastWelcomeMessage();
expect(welcome).not.toBeNull();
expect(welcome?.serverId.length).toBeGreaterThan(0);
expect(typeof welcome?.resumed).toBe("boolean");
const serverInfo = client.getLastServerInfoMessage();
expect(serverInfo).not.toBeNull();
expect(serverInfo?.serverId.length).toBeGreaterThan(0);
await client.close();
}, 15000);
@@ -369,8 +368,8 @@ describe("daemon client E2E", () => {
try {
await client.connect();
const welcome = client.getLastWelcomeMessage();
const voice = welcome?.capabilities?.voice;
const serverInfo = client.getLastServerInfoMessage();
const voice = serverInfo?.capabilities?.voice;
expect(voice).toBeTruthy();
expect(voice?.dictation.enabled).toBe(false);
@@ -243,7 +243,13 @@ async function waitForRelayWebSocketReady(port: number, timeout = 60000): Promis
try {
const payload =
typeof data === "string" ? JSON.parse(data) : data;
if (payload && typeof payload === "object" && (payload as any).type === "welcome") {
if (
payload &&
typeof payload === "object" &&
(payload as any).type === "session" &&
(payload as any).message?.type === "status" &&
(payload as any).message?.payload?.status === "server_info"
) {
if (!pingSent && channelRef) {
pingSent = true;
void channelRef.send(JSON.stringify({ type: "ping" }));
@@ -365,7 +371,13 @@ async function waitForRelayWebSocketReady(port: number, timeout = 60000): Promis
const channel = await createClientChannel(transport, daemonPublicKeyB64, {
onmessage: (data) => {
const payload = typeof data === "string" ? JSON.parse(data) : data;
if (payload && typeof payload === "object" && (payload as any).type === "welcome") {
if (
payload &&
typeof payload === "object" &&
(payload as any).type === "session" &&
(payload as any).message?.type === "status" &&
(payload as any).message?.payload?.status === "server_info"
) {
if (!pingSent && channelRef) {
pingSent = true;
void channelRef.send(JSON.stringify({ type: "ping" }));
@@ -459,7 +459,7 @@ const shouldRun = !process.env.CI;
const helloReady = new Promise<void>((resolve, reject) => {
const timeout = setTimeout(() => {
cleanup();
reject(new Error("Timed out waiting for websocket welcome"));
reject(new Error("Timed out waiting for initial websocket server_info"));
}, 10000);
const onError = (error: Error) => {
@@ -478,8 +478,15 @@ const shouldRun = !process.env.CI;
}
const text = typeof raw === "string" ? raw : raw.toString("utf8");
try {
const parsed = JSON.parse(text) as { type?: string };
if (parsed.type === "welcome") {
const parsed = JSON.parse(text) as {
type?: string;
message?: { type?: string; payload?: { status?: string } };
};
if (
parsed.type === "session" &&
parsed.message?.type === "status" &&
parsed.message.payload?.status === "server_info"
) {
cleanup();
resolve();
}
@@ -282,13 +282,15 @@ async function attachRelayAndHello(params: {
params.socket.emit("message", JSON.stringify(createHelloMessage(params.clientId)));
await Promise.resolve();
expect(params.socket.sent.length).toBeGreaterThan(0);
const welcome = JSON.parse(params.socket.sent[0] as string) as {
const envelope = JSON.parse(params.socket.sent[0] as string) as {
type?: unknown;
resumed?: unknown;
serverId?: unknown;
message?: { type?: unknown; payload?: unknown };
};
expect(welcome.type).toBe("welcome");
return welcome;
expect(envelope.type).toBe("session");
const serverInfo = parseServerInfoStatusPayload(envelope.message?.payload);
expect(envelope.message?.type).toBe("status");
expect(serverInfo).not.toBeNull();
return serverInfo!;
}
async function attachDirectAndHello(params: {
@@ -300,13 +302,15 @@ async function attachDirectAndHello(params: {
params.socket.emit("message", JSON.stringify(createHelloMessage(params.clientId)));
await Promise.resolve();
expect(params.socket.sent.length).toBeGreaterThan(0);
const welcome = JSON.parse(params.socket.sent[0] as string) as {
const envelope = JSON.parse(params.socket.sent[0] as string) as {
type?: unknown;
resumed?: unknown;
serverId?: unknown;
message?: { type?: unknown; payload?: unknown };
};
expect(welcome.type).toBe("welcome");
return welcome;
expect(envelope.type).toBe("session");
const serverInfo = parseServerInfoStatusPayload(envelope.message?.payload);
expect(envelope.message?.type).toBe("status");
expect(serverInfo).not.toBeNull();
return serverInfo!;
}
describe("relay external socket reconnect behavior", () => {
@@ -324,12 +328,11 @@ describe("relay external socket reconnect behavior", () => {
const clientId = "cid-relay-reconnect";
const socket1 = new MockSocket();
const firstWelcome = await attachRelayAndHello({
await attachRelayAndHello({
server,
socket: socket1,
clientId,
});
expect(firstWelcome.resumed).toBe(false);
expect(sessionMock.instances).toHaveLength(1);
const session = sessionMock.instances[0]!;
@@ -338,12 +341,11 @@ describe("relay external socket reconnect behavior", () => {
expect(session.cleanup).not.toHaveBeenCalled();
const socket2 = new MockSocket();
const secondWelcome = await attachRelayAndHello({
await attachRelayAndHello({
server,
socket: socket2,
clientId,
});
expect(secondWelcome.resumed).toBe(true);
expect(sessionMock.instances).toHaveLength(1);
await vi.advanceTimersByTimeAsync(20_000);
@@ -373,50 +375,46 @@ describe("relay external socket reconnect behavior", () => {
await server.close();
});
test("marks hello as resumed when clientId already has a session", async () => {
test("returns server_info when clientId reconnects with existing session", async () => {
const server = createServer();
const clientId = "cid-resume-flag";
const firstSocket = new MockSocket();
const firstWelcome = await attachRelayAndHello({
await attachRelayAndHello({
server,
socket: firstSocket,
clientId,
});
expect(firstWelcome.resumed).toBe(false);
firstSocket.emit("close", 1006, "");
await vi.advanceTimersByTimeAsync(1_000);
const secondSocket = new MockSocket();
const secondWelcome = await attachRelayAndHello({
await attachRelayAndHello({
server,
socket: secondSocket,
clientId,
});
expect(secondWelcome.resumed).toBe(true);
await server.close();
});
test("marks hello as not resumed for new clientIds", async () => {
test("returns server_info for distinct clientIds", async () => {
const server = createServer();
const firstSocket = new MockSocket();
const firstWelcome = await attachRelayAndHello({
await attachRelayAndHello({
server,
socket: firstSocket,
clientId: "cid-new-1",
});
expect(firstWelcome.resumed).toBe(false);
const secondSocket = new MockSocket();
const secondWelcome = await attachRelayAndHello({
await attachRelayAndHello({
server,
socket: secondSocket,
clientId: "cid-new-2",
});
expect(secondWelcome.resumed).toBe(false);
expect(sessionMock.instances).toHaveLength(2);
await server.close();
@@ -456,12 +454,11 @@ describe("relay external socket reconnect behavior", () => {
const clientId = "cid-direct-reconnect";
const socket1 = new MockSocket();
const firstWelcome = await attachDirectAndHello({
await attachDirectAndHello({
server,
socket: socket1,
clientId,
});
expect(firstWelcome.resumed).toBe(false);
expect(sessionMock.instances).toHaveLength(1);
const session = sessionMock.instances[0]!;
@@ -470,12 +467,11 @@ describe("relay external socket reconnect behavior", () => {
expect(session.cleanup).not.toHaveBeenCalled();
const socket2 = new MockSocket();
const secondWelcome = await attachDirectAndHello({
await attachDirectAndHello({
server,
socket: socket2,
clientId,
});
expect(secondWelcome.resumed).toBe(true);
expect(sessionMock.instances).toHaveLength(1);
await vi.advanceTimersByTimeAsync(20_000);
@@ -548,12 +544,12 @@ describe("relay external socket reconnect behavior", () => {
await server.close();
});
test("includes voice capabilities in welcome when speech readiness exists", async () => {
test("includes voice capabilities in initial server_info when speech readiness exists", async () => {
const speechReadiness = createReadySpeechReadinessSnapshot();
const server = createServer({ speechReadiness });
const socket = new MockSocket();
const welcome = await attachRelayAndHello({
const serverInfo = await attachRelayAndHello({
server,
socket,
clientId: "cid-server-info-capabilities",
@@ -566,15 +562,15 @@ describe("relay external socket reconnect behavior", () => {
};
};
};
expect(welcome.version).toBe(TEST_DAEMON_VERSION);
expect(welcome.capabilities?.voice?.dictation?.enabled).toBe(
expect(serverInfo.version).toBe(TEST_DAEMON_VERSION);
expect(serverInfo.capabilities?.voice?.dictation?.enabled).toBe(
speechReadiness.dictation.enabled
);
expect(welcome.capabilities?.voice?.dictation?.reason).toBe("");
expect(welcome.capabilities?.voice?.voice?.enabled).toBe(
expect(serverInfo.capabilities?.voice?.dictation?.reason).toBe("");
expect(serverInfo.capabilities?.voice?.voice?.enabled).toBe(
speechReadiness.realtimeVoice.enabled
);
expect(welcome.capabilities?.voice?.voice?.reason).toBe("");
expect(serverInfo.capabilities?.voice?.voice?.reason).toBe("");
await server.close();
});
+12 -19
View File
@@ -12,7 +12,6 @@ import type { ProjectRegistry, WorkspaceRegistry } from "./workspace-registry.js
import {
type ServerInfoStatusPayload,
type WSHelloMessage,
type WSWelcomeMessage,
WSInboundMessageSchema,
type ServerCapabilityState,
type ServerCapabilities,
@@ -678,17 +677,6 @@ export class VoiceAssistantWebSocketServer {
return pending;
}
private buildWelcomeMessage(params: { resumed: boolean }): WSWelcomeMessage {
return {
type: "welcome",
serverId: this.serverId,
hostname: getHostname(),
version: this.daemonVersion,
resumed: params.resumed,
...(this.serverCapabilities ? { capabilities: this.serverCapabilities } : {}),
};
}
private handleHello(params: {
ws: WebSocketLike;
message: WSHelloMessage;
@@ -735,7 +723,7 @@ export class VoiceAssistantWebSocketServer {
}
existing.sockets.add(ws);
this.sessions.set(ws, existing);
this.sendToClient(ws, this.buildWelcomeMessage({ resumed: true }));
this.sendToClient(ws, this.createServerInfoMessage());
existing.connectionLogger.trace(
{
clientId,
@@ -756,7 +744,7 @@ export class VoiceAssistantWebSocketServer {
});
this.sessions.set(ws, connection);
this.externalSessionsByKey.set(clientId, connection);
this.sendToClient(ws, this.buildWelcomeMessage({ resumed: false }));
this.sendToClient(ws, this.createServerInfoMessage());
connection.connectionLogger.trace(
{
clientId,
@@ -777,13 +765,18 @@ export class VoiceAssistantWebSocketServer {
};
}
private broadcastCapabilitiesUpdate(): void {
this.broadcast(
wrapSessionMessage({
private createServerInfoMessage(): WSOutboundMessage {
return {
type: "session",
message: {
type: "status",
payload: this.buildServerInfoStatusPayload(),
})
);
},
};
}
private broadcastCapabilitiesUpdate(): void {
this.broadcast(this.createServerInfoMessage());
}
private bindSocketHandlers(ws: WebSocketLike): void {
-11
View File
@@ -2388,15 +2388,6 @@ export const WSHelloMessageSchema = z.object({
.optional(),
})
export const WSWelcomeMessageSchema = z.object({
type: z.literal('welcome'),
serverId: z.string(),
hostname: z.string().nullable(),
version: z.string().nullable(),
resumed: z.boolean(),
capabilities: ServerCapabilitiesSchema.optional(),
})
export const WSRecordingStateMessageSchema = z.object({
type: z.literal('recording_state'),
isRecording: z.boolean(),
@@ -2423,14 +2414,12 @@ export const WSInboundMessageSchema = z.discriminatedUnion('type', [
export const WSOutboundMessageSchema = z.discriminatedUnion('type', [
WSPongMessageSchema,
WSWelcomeMessageSchema,
WSSessionOutboundSchema,
])
export type WSInboundMessage = z.infer<typeof WSInboundMessageSchema>
export type WSOutboundMessage = z.infer<typeof WSOutboundMessageSchema>
export type WSHelloMessage = z.infer<typeof WSHelloMessageSchema>
export type WSWelcomeMessage = z.infer<typeof WSWelcomeMessageSchema>
// ============================================================================
// Helper functions for message conversion