3 ms·
> The primary advantages center around the ergonomics of concurrency. That is how most "frameworks" or those that try to copy this aspect from Erlang see it. B
by rdtsc 12y ago
> The primary advantages center around the ergonomics of concurrency.
That is how most "frameworks" or those that try to copy this aspect from Erlang see it. But the trick is in Erlang this actor pattern is used for fault-tolerance just as much. That was the goal right alongside concurrency initially. (The third one was priority of low latency responses, I believe).
That fault tolerance is harder to copy and that is why most libraries and frameworks give up and copy the "class object+thread+queue" and call it "We have a fast Erlang now".
The closest to get to something like Erlang fault tolerance wise is to use OS processes and IPC (via ZMQ), with some serialization. But it would be hard to run 2M of those on a reasonable machine.
Plus Erlang is not just the language (which is rather small and simple) but the whole set of helper libraries and tools. Including a distributed database, distributed application controllers. Support for rpc and so on. Supervisor patterns you can use etc.