Hi,

this is coming from another language, but maybe is also a solution here. The problem with polling is that is scales poorly and can result either in high load or long delays (it all comes down to your application requirements really). In our case we have a high volume and low latency requirements, so polling isn’t an option.

In one project, we use a Faye server and JS libraries to affect the client state. The server is standalone, so you can just run it as a separate service (i.e. it doesn’t matter much in which language it is really; it is a Ruby application); it provides a basic WebSocket connection with a simple PubSub-protocol on top of it.

A JS library on the client side connects to the websocket, registers for a bunch of channels and waits (e.g. a channel like /user/some-magic-hash for a user-specific channel)

On the server side you can use a fairly straightforward API to push messages to faye to inform the client to do something. Depending on how much effort you put into the client you can more or less push arbitrary JS oder DOM changes from the server side then.

In your example it might be that when the button is pressed, the server pushes an “enable-form-xyz” message to the user’s channel. The client on the second computer, which is connected to the same channel, receives an “enable-form-xyz” and can use some simple JS manipulation to enable form fields. Of course you should always do additional server-side validation in case the second computer enters new data and such.

Isn’t quite as simple to setup but has some scaling advantages and provides fairly instant feedback. The protocol used is called Bayeux and I think there are other implementations, Faye for our purposes is just a very simple standalone service to use here.

Greetings

Dennis

Links:

• https://faye.jcoglan.com/ Faye Server/Client