Smoke test for tracing infrastructure in PyAuto

chrome/test/functional/tracing/tracing_base.py

Index: chrome/test/functional/tracing/tracing_base.py
diff --git a/chrome/test/functional/tracing/tracing_base.py b/chrome/test/functional/tracing/tracing_base.py
new file mode 100644
index 0000000000000000000000000000000000000000..deb533a9f600c13dc2b8bad6d34ad7cecacb3785
--- /dev/null
+++ b/chrome/test/functional/tracing/tracing_base.py
@@ -0,0 +1,51 @@
+#!/usr/bin/env python
+# Copyright (c) 2011 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.
+
+import unittest
+
+import pyauto_tracing
+import pyauto
+
+from timeline_model import TimelineModel
+
+class TracingTestBase(pyauto.PyUITest):

[nduca, 2012/07/18 05:51:35]

Typically, python code names the file after the class it contains. So this file
would be TracingTestBase.

Though it looks like pyauto uses the conventin XXXTest.

So this would just be TracingTest, inside tracing_test.py

+  def setUp(self):
+    super(TracingTestBase, self).setUp()
+    self._trace_win = self.GetBrowserWindowCount()
+    self.OpenNewBrowserWindow(False)
+    self.NavigateToURL('chrome://tracing', self._trace_win)
+    self.ExecuteJavascript("""
+      tracingController.addEventListener('traceEnded', function() {

[nduca, 2012/07/18 05:51:35]

This feels like it could be in the shim code.

E.g. have the shim define an export "recordTrace."

+        window.pyauto_timeline_model = new tracing.TimelineModel();

[nduca, 2012/07/18 05:51:35]

Instead of creating a timeline model directly, create a shim that wraps or
subclasses the timeline model. This gives you a clean place where you can add
functions to the TimelineModel that are specific to its use from python.

[nduca, 2012/07/18 05:51:35]

We use window.fooBar convention in js for globals. Its enough to drive you mad
:)

http://google-styleguide.googlecode.com/svn/trunk/javascriptguide.xml#Naming

+        var events = [tracingController.traceEvents_];

[Russ Harmon, 2012/07/14 11:58:45]

I wasn't able to use tracingController.traceData as that only gets set when you
read from a file.

+        if (tracingController.supportsSystemTracing)
+          events.push(tracingController.systemTraceEvents_);
+        window.pyauto_timeline_model.importTraces(events);
+        window.domAutomationController.send('');
+      });
+      window.domAutomationController.send('');
+    """, 0, self._trace_win)
+
+  def tearDown(self):
+    self.CloseBrowserWindow(self._trace_win)
+    self._trace_win = None

[nduca, 2012/07/18 05:51:35]

Is this conventional, to assign the trace_win to none here? Its fine if other
tests do this, but it just smells funny to me.

+    super(TracingTestBase, self).tearDown()
+
+  def _BeginTracing(self, system_tracing=True):

[nduca, 2012/07/18 05:51:35]

Where's this get called from?

+    # TODO(nduca) Make the chrome tracing ui easier to automate
+    self.ExecuteJavascript("""
+      tracingController.beginTracing(
+        // Warning, possible javascript injection here!
+        tracingController.supportsSystemTracing ? %s : false
+      );
+      window.domAutomationController.send('');
+    """ % ("true" if system_tracing else "false"), 0, self._trace_win)
+
+  def _EndTracing(self):

[nduca, 2012/07/18 05:51:35]

Where's this called from?

+    # TODO(nduca) Make the chrome tracing ui easier to automate
+    self.ExecuteJavascript("""
+      tracingController.endTracing();
+    """, 0, self._trace_win)
+    return TimelineModel(self, "window.pyauto_timeline_model")