Discussion summary: Conversational Context & Concurrency

Speaker: Tony Garnock-Jones

(Tony’s notes are excellent, this is just a summary of the discussion.)

M: people in the 70s were very concerned about syntax

With monitors (think: Java or POSIX threads) you have to be careful to propagate exceptions from child threads to parent threads.

M: its mind-boggling how many reasoned Racket programmers forget to communicate events to other threads. Very common mistake.

?: how do you GC dead processes in the tuple-space world? T: it’s implementation-dependent

T: π-calculus people love to encode variants of the calculus into each other

Actors were invented for artificial intelligence, the original paper is VERY interesting. Literally everything was going to be an actor. (Hewitt Bishop Steiger 1973, see Tony’s notes)

W: The actor model paper came out just after the first laser printers. There are some great fonts in the paper!

M: when you have a huge space like this — models of concurrent programming — your first job as a researcher is to do what Tony did and determine a criteria for evaluating and comparing the different approaches.