Accept iterator as argument to ewma

This avoids an intermediate call to collect, providing a small
performance increase.

src/socket.rs

@@ -727,7 +727,7 @@ impl UtpSocket {
     /// weighted moving average filter with smoothing factor 0.333 over the
     /// current delays in the current window.
     fn filtered_current_delay(&self) -> i64 {
-        let input = self.current_delays.iter().map(|x| x.difference).collect();
+        let input = self.current_delays.iter().map(|x| x.difference);
         ewma(input, 0.333) as i64
     }
 

src/util.rs

@@ -9,9 +9,9 @@ pub fn now_microseconds() -> u32 {
 
 /// Calculate the exponential weighted moving average for a vector of numbers, with a smoothing
 /// factor `alpha` between 0 and 1. A higher `alpha` discounts older observations faster.
-pub fn ewma<T: ToPrimitive>(samples: Vec<T>, alpha: f64) -> f64 {
-    samples.iter().fold(samples.first().map_or(0.0, |v| v.to_f64().unwrap()),
-                        |avg, sample| alpha * sample.to_f64().unwrap() + (1.0 - alpha) * avg)
+pub fn ewma<T: ToPrimitive, I: Iterator<Item=T>>(mut samples: I, alpha: f64) -> f64 {
+    let first = samples.next().map_or(0.0, |v| v.to_f64().unwrap());
+    samples.fold(first, |avg, sample| alpha * sample.to_f64().unwrap() + (1.0 - alpha) * avg)
 }
 
 /// Returns the absolute difference between two integers.
@@ -32,19 +32,19 @@ mod test {
     fn test_ewma_empty_vector() {
         let empty: Vec<u32> = vec!();
         let alpha = 1.0/3.0;
-        assert_eq!(ewma(empty, alpha), 0.0);
+        assert_eq!(ewma(empty.iter().map(|x| *x), alpha), 0.0);
     }
 
     #[test]
     fn test_ewma_one_element() {
         let input = vec!(1u32);
         let alpha = 1.0/3.0;
-        assert_eq!(ewma(input, alpha), 1.0);
+        assert_eq!(ewma(input.iter().map(|x| *x), alpha), 1.0);
     }
 
     #[test]
     fn test_exponential_smoothed_moving_average() {
-        let input = (1u32..11).collect();
+        let input = 1u32..11;
         let alpha = 1.0/3.0;
         let expected = [1.0, 4.0/3.0, 17.0/9.0,
         70.0/27.0, 275.0/81.0, 1036.0/243.0, 3773.0/729.0, 13378.0/2187.0,