Care to back that up w/ code? It is certainly possible, even likely. But you need to do it w/ processes (not threads). And you need to define message passing. Using shared memory or mapped memory probably doesn’t count. TCP/IP is probably a must using ASN.1 or UBF over XML is prefered.

Write us up some code and show us the results! Then show us the number of lines of code you wrote including the ASN.1 or UBF functions. Then compare the time and effort it took. Take the ratio of you LOCs vs. 31 in ring.erl.

Turing proved that any problem solved in 1 language can be solved in another (or something like that) … but it doesn’t mean you *should* solve certain problems in YOUR language. There is a freedom of choice.

Otherwise, it’s not really fair towards platforms that use OS threads, because OS threads have clear disadvantages, but allow sharing the load over multiple CPUs.

-module(ring).
-export([go/2]).

go(N, M) ->
FPid = spawn(fun() -> initfirst() end),
SPid = spawn(fun() -> initrest(N, 2, M, FPid) end),
register(master, self()),
FPid ! {start, SPid, M, hello},
wait_done(FPid).

wait_done(FPid) ->
receive
{finished, FPid} -> unregister(master)
end.

initfirst() ->
receive
{start, NextP, M, Msg} ->
NextP ! {Msg},
loop(M, NextP, self())
end.

initrest(N, N, M, FPid) -> loop (M, FPid, FPid);
initrest(N, J, M, FPid) ->
NextId = spawn(fun()->initrest(N, J+1, M, FPid) end),
loop (M, NextId, FPid).

loop(M, NextP, FPid) when M > 0 ->
receive
{Msg} ->
NextP!{Msg}
end,
loop(M-1, NextP, FPid);
loop(0, _, FPid) when FPid == self() -> master!{finished, self()};
loop(0, _, _) -> true.