I know how to model a state machine with a flow chart, it make sense.
But I’m designing an API to control websockets and manage a web websocket service via HTTP similar to pushpin. So I have at least 4 state machines. But then there’s the events and commands that get sent around that change state and I’m at a complete loss of how to model that.
I can muddle through in my head but, I want to prove it out somehow. Anyone have any experience modeling state in a distributed system?
I have at least 4 types of actors.
- API Gateway and it's connection state
- The lambda’s it executes when connections have activity (open, message, disconnect)
- Your application getting a http request to notify you of events and get commands
- and the commands you send to the control api anytime
The last three could have any number of them running at a time
And then I got quirks of api gateway which suck for application developers (one reason I'm not using it directly and building this service)
You have an open callback, and you can reject the connection or accept it, by the return of the callback. So you can't actually do any work (eg, sending messages) in this callback and you have to do the work after it returns. But how do you know if it's returned?
Francis 🏴☠️ Gulotta
Jed Schmidt