wsApp.publish or ws.send?

[kvalood]

Hi all!
What will be more productive when sending 100 messages to 1000 users per second?

wsApp.publish('topic', message)

or

clients = new Map() // or object
clietns.forEach(u => {
    u.send(message)
})

[e3dio]

Publish is much better. Not only will it iterate over 1000 clients faster, but the main feature of Publish is it automatically groups outgoing messages by socket before sending out, if you publish 100 different messages to some subset of 1000 clients it will be 100 times more efficient and faster than looping over and sending each message individually, because it can send all messages to a client at same time

[kvalood]

@e3dio I changed the original question a bit. Please read.

[e3dio]

Still have no idea what you are asking. Subscribe/Publish by topic is very easy and straightforward, it is not rocket science

[e3dio]

How can one client subscribe twice to the same data source?

This is the only question you asked and it makes zero sense, unless you want duplicate data, which still makes no sense. Need to write a question that makes sense

[kvalood]

This is the only question you asked and it makes zero sense, unless you want duplicate data, which still makes no sense. Need to write a question that makes sense

Duplicate sending data from server to client because client subscribed multiple times

Doesn't make sense, because on the client side, you can receive data once, and output it several times to different places.

[e3dio]

Duplicate sending data from server to client because client subscribed multiple times

You should probably consider re-writing your client code to not subscribe to the same data source multiple times, avoid sending duplicate data. Just a hint ;)

[uNetworkingAB]

Hi all! What will be more productive when sending 100 messages to 1000 users per second?

wsApp.publish('topic', message)

or

clients = new Map() // or object
clietns.forEach(u => {
    u.send(message)
})

wsApp.publish('topic', message) is made for this, so use it. But also make your own benchmarks. The follow up question I don't understand, you use loosely defined, or simply undefined, terminology.