Roll Observatory packages and add a roll script

packages/observe/benchmark/setup_observation_benchmark_base.dart

+// Copyright (c) 2013, the Dart project authors.  Please see the AUTHORS file
+// for details. All rights reserved. Use of this source code is governed by a
+// BSD-style license that can be found in the LICENSE file.
+library observe.test.benchmark.setup_observation_benchmark_base;
+
+import 'dart:async';
+import 'dart:html';
+import 'package:observe/observe.dart';
+import 'package:benchmark_harness/benchmark_harness.dart';
+
+abstract class SetupObservationBenchmarkBase extends BenchmarkBase {
+  /// The number of objects to create and observe.
+  final int objectCount;
+
+  /// The current configuration.
+  final String config;
+
+  /// The objects we want to observe.
+  List<Observable> objects;
+
+  SetupObservationBenchmarkBase(String name, this.objectCount, this.config)
+      : super(name);
+
+  /// Subclasses should use this method to return an observable object to be
+  /// benchmarked.
+  Observable newObject();
+
+  /// Subclasses should override this to do anything other than a default change
+  /// listener. It must return either a StreamSubscription or a PathObserver.
+  newObserver(obj) => obj.changes.listen((_) {});
+
+  /// Set up each benchmark by creating all the objects.
+  @override
+  void setup() {
+    objects = [];
+    while (objects.length < objectCount) {
+      objects.add(newObject());
+    }
+  }
+
+  /// Tear down each the benchmark and remove all listeners.
+  @override
+  void teardown() {
+    while (objects.isNotEmpty) {
+      var obj = objects.removeLast();
+      if (obj.hasObservers || (obj is ObservableList && obj.hasListObservers)) {
+        window.alert('Observers leaked!');
+      }
+    }
+    objects = null;
+  }
+
+  /// Run the benchmark by creating a listener on each object.
+  @override
+  void run() {
+    for (var object in objects) {
+      var observer = newObserver(object);
+
+      // **Note:** This is different than the JS implementation. Since run can
+      // be called an arbitrary number of times between [setup] and [teardown],
+      // we clean up all observers as we go. This means we are measuring both
+      // the setup and teardown of observers, versus the setup only in the
+      // JS benchmark. Not cleaning these up ends up giving `oh snap` errors.
+      if (observer is StreamSubscription) {
+        observer.cancel();
+      } else if (observer is PathObserver) {
+        observer.close();
+      } else {
+        throw 'Unknown observer type ${observer.runtimeType}. Only '
+            '[PathObserver] and [StreamSubscription] are supported.';
+      }
+    }
+  }
+}