Add some basic interfaces that will be used for the test runner.

lib/src/live_test.dart

Index: lib/src/live_test.dart
diff --git a/lib/src/live_test.dart b/lib/src/live_test.dart
new file mode 100644
index 0000000000000000000000000000000000000000..646f5de169a18774652b45126168cc19b77bdcaa
--- /dev/null
+++ b/lib/src/live_test.dart
@@ -0,0 +1,104 @@
+// Copyright (c) 2015, 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 unittest.live_test;
+
+import 'dart:async';
+
+import 'state.dart';
+import 'suite.dart';
+import 'test.dart';
+
+/// A runnable instance of a test.
+///
+/// This is distinct from [Test] in order to keep [Test]. Running a test
+/// requires state, and [LiveTest] provides a view of the state of the test as
+/// it runs.
+///
+/// If the state changes, [state] will be updated before [onStateChange] fires.
+/// Likewise, if an error is caught, it will be added to [errors] before being
+/// emitted via [onError]. If an error causes a state change, [onStateChange]
+/// will fire before [onError]. If an error or other state change causes the
+/// test to complete, [onComplete] will complete after [onStateChange] and
+/// [onError] fire.
+abstract class LiveTest {
+  /// The suite within which this test is being run.
+  Suite get suite;
+
+  /// The running test.
+  Test get test;
+
+  /// The current state of the running test.
+  ///
+  /// This starts as [Status.pending] and [Result.success]. It will be updated
+  /// before [onStateChange] fires.
+  ///
+  /// Note that even if this is marked [Status.complete], the test may still be
+  /// running code asynchronously. A test is considered complete either once it
+  /// hits its first error or when all [expectAsync] callbacks have been called
+  /// and any returned [Future] has completed, but it's possible for further
+  /// processing to happen, which may cause further errors. It's even possible
+  /// for a test that was marked [Status.complete] and [Result.success] to be
+  /// marked as [Result.error] later.
+  State get state;
+
+  /// Returns whether this test has completed.
+  ///
+  /// This is equivalent to [state.status] being [Status.complete].
+  ///
+  /// Note that even if this returns `true`, the test may still be
+  /// running code asynchronously. A test is considered complete either once it
+  /// hits its first error or when all [expectAsync] callbacks have been called
+  /// and any returned [Future] has completed, but it's possible for further
+  /// processing to happen, which may cause further errors.
+  bool get isComplete => state.status == Status.complete;
+
+  // A stream that emits a new [State] whenever [state] changes.
+  //
+  // This will only ever emit a [State] if it's different than the previous
+  // [state]. It will emit an event after [state] has been updated. Note that
+  // since this is an asynchronous stream, it's possible for [state] not to
+  // match the [State] that it emits within the [Stream.listen] callback.
+  Stream<State> get onStateChange;
+
+  /// An unmodifiable list of all errors that have been caught while running
+  /// this test.
+  ///
+  /// This will be updated before [onError] fires. These errors are not
+  /// guaranteed to have the same types as when they were thrown; for example,
+  /// they may need to be serialized across isolate boundaries. The stack traces
+  /// will be [Chain]s.
+  List<AsyncError> get errors;
+
+  /// A stream that emits a new [AsyncError] whenever an error is caught.
+  ///
+  /// This will be emit an event after [errors] is updated. These errors are not
+  /// guaranteed to have the same types as when they were thrown; for example,
+  /// they may need to be serialized across isolate boundaries. The stack traces
+  /// will be [Chain]s.
+  Stream<AsyncError> get onError;
+
+  /// A [Future] that completes once the test is complete.
+  ///
+  /// This will complete after [onStateChange] has fired, and after [onError]
+  /// has fired if the test completes because of an error. It's the same as the
+  /// [Future] returned by [run].
+  ///
+  /// Note that even once this completes, the test may still be running code
+  /// asynchronously. A test is considered complete either once it hits its
+  /// first error or when all [expectAsync] callbacks have been called and any
+  /// returned [Future] has completed, but it's possible for further processing
+  /// to happen, which may cause further errors.
+  Future get onComplete;
+
+  /// Signals that this test should start running as soon as possible.
+  ///
+  /// A test may not start running immediately for various reasons specific to
+  /// the means by which it's defined. Until it starts running, [state] will
+  /// continue to be marked [Status.pending].
+  ///
+  /// This returns the same [Future] as [onComplete]. It may not be called more
+  /// than once.
+  Future run();
+}