bench: Add threadring shootout benchmark

src/test/bench/shootout-threadring.rs

@@ -0,0 +1,50 @@
+// Based on threadring.erlang by Jira Isa
+use std;
+
+const n_threads: int = 503;
+
+fn start(+token: int) {
+    import iter::*;
+
+    let p = comm::port();
+    let ch = iter::foldl(bind int::range(2, n_threads + 1, _),
+                         comm::chan(p)) { |ch, i|
+        // FIXME: Some twiddling because we don't have a standard
+        // reverse range function yet
+        let id = n_threads + 2 - i;
+        let {to_child, _} = task::spawn_connected::<int, int> {|p, _ch|
+            roundtrip(id, p, ch)
+        };
+        to_child
+    };
+    comm::send(ch, token);
+    roundtrip(1, p, ch);
+}
+
+fn roundtrip(id: int, p: comm::port<int>, ch: comm::chan<int>) {
+    while (true) {
+        alt comm::recv(p) {
+          1 {
+            std::io::println(#fmt("%d\n", id));
+            ret;
+          }
+          token {
+            #debug("%d %d", id, token);
+            comm::send(ch, token - 1);
+            if token <= n_threads {
+                ret;
+            }
+          }
+        }
+    }
+}
+
+fn main(args: [str]) {
+    let token = if vec::len(args) < 2u {
+        1000
+    } else {
+        int::from_str(args[1])
+    };
+
+    start(token);
+}