Track all extension frames in ProcessManager, inspect extensionoptions

chrome/test/data/extensions/api_test/messaging/connect_nobackground/contentscript.js

Index: chrome/test/data/extensions/api_test/messaging/connect_nobackground/contentscript.js
diff --git a/chrome/test/data/extensions/api_test/messaging/connect_nobackground/contentscript.js b/chrome/test/data/extensions/api_test/messaging/connect_nobackground/contentscript.js
new file mode 100644
index 0000000000000000000000000000000000000000..f590996543c8815f7d23a75004a23179630566ca
--- /dev/null
+++ b/chrome/test/data/extensions/api_test/messaging/connect_nobackground/contentscript.js
@@ -0,0 +1,11 @@
+// Copyright (c) 2015 The Chromium Authors. All rights reserved.

[Devlin, 2015/10/30 01:21:43]

nit: no (c)

+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+var port = chrome.runtime.connect();
+port.postMessage('Hello from content script');
+port.disconnect();
+
+chrome.runtime.onMessage.addListener(function(message, sender, sendResponse) {

[Devlin, 2015/10/30 01:21:43]

this seems like a subtle race condition - can we register the listener before we
disconnect?

[robwu, 2015/10/30 10:15:49]

This can't race because the message has to go to another process via IPC. If a
race would occur, then it means that the JS engine was blocked while waiting for
a reply, and that is something that should cause a test to fail.

[Devlin, 2015/10/30 16:03:51]

Generally in testing, though, we don't want to rely on implementation details
like that.  What if one day all the messaging code lived in the renderer, and
could be synchronous?

+  sendResponse('Reply here');

[Devlin, 2015/10/30 01:21:43]

Is it worth verifying the message?  Probably not necessary, but it's a super
cheap assert, so maybe better to do it.

+});