Dismiss multiple error dialogs when running tests

tools/telemetry/telemetry/core/backends/chrome/android_browser_backend.py

Index: tools/telemetry/telemetry/core/backends/chrome/android_browser_backend.py
diff --git a/tools/telemetry/telemetry/core/backends/chrome/android_browser_backend.py b/tools/telemetry/telemetry/core/backends/chrome/android_browser_backend.py
index 0fba7e7602c54de55a0896c9470adcc9577d3b6b..9efe747f0f024aebb2f1357408651772195f94fd 100644
--- a/tools/telemetry/telemetry/core/backends/chrome/android_browser_backend.py
+++ b/tools/telemetry/telemetry/core/backends/chrome/android_browser_backend.py
@@ -273,7 +273,11 @@ class AndroidBrowserBackend(chrome_browser_backend.ChromeBrowserBackend):
       # If we have no existing tabs start with a blank page since default
       # startup with the NTP can lead to race conditions with Telemetry
       url = 'about:blank'
-    self._adb.device().old_interface.DismissCrashDialogIfNeeded()
+    # Dismiss any error dialogs. Limit the number in case we have an error loop
+    # or we are failing to dismiss.
+    for _ in xrange(10):
+      if not self._adb.device().old_interface.DismissCrashDialogIfNeeded():

[tonyg, 2014/08/05 14:34:01]

Are there any users of this API that don't want all dialogs to be dismissed? I
wonder whether the for-loop should be inside it instead of around it?

[aberent, 2014/08/05 14:41:40]

I considered this, but code in test_runner.py (see above) wants to check whether
the dialog comes from the test package, so it is easiest to leave the loop
outside. The alternative would be to return a list of the packages which had had
error dialogs.

+        break
     self._adb.device().StartActivity(
         intent.Intent(package=self._backend_settings.package,
                       activity=self._backend_settings.activity,