I do not agree with the contents of this article. Behaviors are possible because of the underlying architecture of the system. Behaviors are not interfaces – t

I do not agree with the contents of this article. Behaviors are possible because of the underlying architecture of the system. Behaviors are not interfaces – they are more like abstract object in a language like Java so they implement basic and self-contained functionalities that are hidden behind a collaboration interface, but they couldn’t do much without the underlying infrastructure that makes sure that every process is totally separate from other processes, the that all processes can be safely closed without leaking memory, or resources, and that you cannot just share a rogue pointer between two different processes. What Joe did in his thesis is to show you how you can build reliable systems (and up point, reliable distribuited systems) by using a given set of Lego blocks. The reason why you need the erlang vm to implement something like that appropriately – and that you cannot do that fully on a different VM – is that without the underlying plumbing, supervision trees would be leaky - in Java, you cannot kill a thread that is holding up to resources and hope that everything will always go well, And do not have ways to monitor different processes.