Comments on: Running rings around Erlang

By: Michael

Michael — Fri, 27 Feb 2009 05:27:18 +0000

I also tried this problem last night, here is my solution.

http://sudothinker.com/2009/2/27/erlang-exercises-interaction-between-processes-concurrency-ring-code

I took it one step further and made sure each node also knew the node before it so it knew where to accept the message from, that way we can be sure that the message is making its correct path around the ring. Might be overkill for this type of problem, but in a bigger system it would make sense as the processes could get messages from anyone and we might not want to send them around the ring.

Thanks for sharing!

By: Alex

Alex — Fri, 02 Jan 2009 17:56:34 +0000

@codist – I’m running on a MacBook Pro with a 2.4 GHz Intel Core 2 Duo and 4 GB of RAM just for the record.

@pat – many thanks for the reference! I knew there was a more clever way to do it.

After sleeping on this blog, it also occurred to me that rather than send a stop message around the ring I could use linked processes and just have the ring die off in reverse based on trapping exit signals. I might give that a try later.

By: pat

pat — Fri, 02 Jan 2009 17:42:43 +0000

See also the Erlang thread-ring implementation at the Computer Language Benchmark Game (Shootout). It uses lists:foldl() to spawn and connect the processes in one go, with self() also used as one process in the ring.

By: codist

codist — Fri, 02 Jan 2009 17:09:46 +0000

A year ago I did a similar thing in Java on a 4 processor PC. Best I could get was 18,000 messages per second.

By: Jonas

Jonas — Fri, 02 Jan 2009 14:51:20 +0000

Great post. Fun example.
Good to see you ramping up on Erlang. I really like it, only wish I could get paid to hack it :)
Looking forward to more posts like this.