I have been checking the documentation and the code but I still don't have a solid concept of how they behave exactly.

I have a codebase that only has at most 1 Socket per room. This is so I can easily emit messages sent to the socket.io server via an internal HTTP api to the user they are meant for (the room name is the userid and its autojoined on connect).

The behavior that would fit my usecase is:

• If the HTTP api gets a message for a user that has never connected the message should be dropped
• If a client disconnects it should also leave the room
• For an emit_with_ack, I'd like to wait for any client to become available to recieve the message before it is dropped for 10 seconds. I do not want to loose the message if the client disconnected the instant before the event was transmitted, but I do not want to keep messages in memory indefinitely either

I am uncertain which of these are default behavior and which have to be manually implemented.

Messages aren't produced unless a client is connected, this is moreso to conserve memory (by avoiding queueing messages that will never be recieved) and ensure consitency (to avoid race conditions)