middleware vs running after connection

[TriStarGod]

Hi,
I'm working with socket.io for the first time to setup auth.
Whats the difference between using a middleware vs running a function after connection?

Ref: https://socket.io/docs/v4/middlewares/

// middleware
io.use((s, next) => {
 if (isAuthed(s.request)) {
  next();
 } else {
  next(new Error("invalid"));
 }
});

vs

// after connection
io.of("/").on("connection", (s) => {
 if (!isAuthed(s.request)) throw new Error("invalid");
 ...
});

I asked this question on stackoverflow but no one answered.

[darrachequesne]

Hi! Technically, there is no difference, the underlying connection (HTTP long-polling or WebSocket) is already established.

That being said, middlewares are the best place to check if the user is authenticated/has sufficient permissions (before the connection at the Socket.IO level). I guess it makes more sense from a client point of view:

• middleware: the client receives a connect_error event upon failure
• in the connection handler: the client first receives a connect event and then a disconnect event

[TriStarGod]

Thanks! That makes a difference to my frontend setup.