I don't yet know how to efficiently answer questions about the Swift language, so forgive if the answer is obvious.
Erlang/Elixir have a fault-intolerance library built on top of their version of actors. https://en.wikipedia.org/wiki/Open_Telecom_Platform It involves "supervision trees”, where supervisors control the lifespans of actors they supervise, react to unexpected failures, etc.
@dgregor79@marick The isolation model implemented in erlang heavily relies on how a "(erlang) process" (aka. actor), is implemented in the BEAM VM. They get their separate mini-heaps and truly "share nothing". This way one actor can "die" and not impact others.
Swift's actors (as well as Orleans, Akka etc), don't have such strong isolation model, and the actors DO share a common heap, and therefore could technically corrupt shared data. 1/2
@nighthawk@marick mini clarification: distributed actors are stable and part of the language. The cluster library didn’t hit 1.0 yet but it’s been used in production in some places.
You could also write your own IPC system using the language’s distributed actors :)
But yeah I answered in another toot that we don’t really have the same way to deal with dislocation as the beam vm
Add comment