Allow webkitSpeechRecognition work from an app's background page.

chrome/test/data/extensions/platform_apps/speech/background_page/test.js

Index: chrome/test/data/extensions/platform_apps/speech/background_page/test.js
diff --git a/chrome/test/data/extensions/platform_apps/speech/background_page/test.js b/chrome/test/data/extensions/platform_apps/speech/background_page/test.js
new file mode 100644
index 0000000000000000000000000000000000000000..fb2407c1b4e685e9fea5c21d30987544304b63a1
--- /dev/null
+++ b/chrome/test/data/extensions/platform_apps/speech/background_page/test.js
@@ -0,0 +1,24 @@
+// Copyright 2013 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+chrome.app.runtime.onLaunched.addListener(function() {
+  chrome.test.runTests([
+    function testWebkitSpeechRecognition() {
+      var succeeded = false;
+
+      var r = new webkitSpeechRecognition();
+      r.onerror = function(e) {
+        if (succeeded) {
+          return;

[not at google - send to devlin, 2013/09/21 04:13:55]

so.. you shouldn't need this anymore right? if I understood correctly? since it
won't fail?

[lazyboy, 2013/09/21 06:07:19]

This is just a safeguard. A request can start and then return onerror because
for tests there won't be any speech recognition servers. I tried to use
mock_speech, but that does too much mocking and bypasses the manifest check.
So for my test case, an onstart at the beginning is good enough to say the test
is exercising what I want.

+        }
+        chrome.test.fail();
+      };
+      r.onstart = function() {
+        succeeded = true;
+        chrome.test.succeed();
+      };
+      r.start();
+    }
+  ]);
+});