As I begin my long trek towards becoming an expert in topics of concurrency, my first stop is going to be Actor Models. Erlang was one of the first languages to really make me consider the value of message passing outside of clunky models such as MPI and to think about how languages could build concurrency right in at a fundamental level. Unfortunately, I never had much excuse before now to sit and play with Erlang as much as I would have liked. So, starting slowly over the next couple of weeks (which are littered with distractions from coursework) and then picking up heavily over Spring Break, I'm going to be reading plenty of papers on Actors and The Inheritance Anomaly as well as trying to design some non-trivial, concurrent program in Erlang (perhaps some sort of YAWS style webserver). Expect a healthy decompressing of Carl Hewitt's foundational "A Universal Modular Actor Formalism for Artificial Intelligence" sometime in the coming week.
Historically, Actors have been used in distributed systems, but I feel like the semantics and the fundamental idea may have value elsewhere. We'll see.
No comments:
Post a Comment