ceil/floor/truncate/round return integers.

runtime/lib/double.dart

Index: runtime/lib/double.dart
diff --git a/runtime/lib/double.dart b/runtime/lib/double.dart
index decfb570389fea5f78f88785634a8c34bb868fbf..29247adb02c05c794bc74c09ba5d709fd5be3ce1 100644
--- a/runtime/lib/double.dart
+++ b/runtime/lib/double.dart
@@ -100,10 +100,15 @@ class _Double implements double {
     return this < 0.0 ? -this : this;
   }
 
-  double round() native "Double_round";
-  double floor() native "Double_floor";
-  double ceil () native "Double_ceil";
-  double truncate() native "Double_truncate";
+  int round() => roundToDouble().toInt();
+  int floor() => floorToDouble().toInt();
+  int ceil () => ceilToDouble().toInt();
+  int truncate() => truncateToDouble().toInt();
+
+  double roundToDouble() native "Double_round";
+  double floorToDouble() native "Double_floor";
+  double ceilToDouble() native "Double_ceil";
+  double truncateToDouble() native "Double_truncate";

[srdjan, 2013/02/26 16:35:26]

Why not use roundToInt, and leave original semantic in? It is expected to that
every operation with double as receiver returns a double  (pow, abs, +, - ...)
unless explicitly said (toInt(), isNegative ..). 

[floitsch, 2013/02/26 17:44:35]

That's already not the case: 5.2 ~/ 2.0 => 2

 
   num clamp(num lowerLimit, num upperLimit) {
     if (lowerLimit is! num) throw new ArgumentError(lowerLimit);