In S1S2 and webappstartup tests, we try to convert the absolute times for the Throbber start|stop, chrome browser startup and WEBAPP STARTUP COMPLETE to times relative to the start of Fennec by determining a "start time" which is substracted from the absolute values found in logcat. The preferred method is to use the Start proc for fennec but this is not always available depending on the Android version and the Phone manufacturer. We should prefer Start proc if it is available and if it is not should not restrict the fallback to the first message containing Gecko but should use messages which contain gecko as well. We should also log the choices we make so that we can investigate issues which may arise.

https://github.com/mozilla/autophone/commit/b06098c145865869616fc0d0fe8ad4005b6993a3 I'm landing this with a self review in order to finish the deployment. jmaher, please let me know if there is anything you would like me to change.

Comment on attachment 8669396 [details] [diff] [review] bug-1211228-v1.patch Review of attachment 8669396 [details] [diff] [review]: ----------------------------------------------------------------- it is sad these are almost identical bits of code for s1s2test and webappstartup, but there is little need to refactor, especially if webappstartup might not be used in the future depending on the decisions around webapps.