olivierli / chrome_safari_power Goto Github PK

Right now it depends on chrome closing itself with a sudden abort inserted in the code.

It should use the driver scripts to run for a specific amount of time.

It's not perfect because the user can always end the session manually but you'd at least want to verify that the app creates a session by default.

This should be done from the command line. Probably should be extracted in a function to be used from compare_chromes.sh?

Or more probably these two should be joined.

I've found it creating a substantial number of idle wake-ups.

It tends to be higher than other readings and not mean much because it includes startup.

Maybe just kill and not kill -9?

Otherwise this can keep youtube from starting the video.

There should be a small standalone program that can be run next to the browser and logs the battery discharge.

This allows calculating how long the full battery would last if executing the scenario over and over.

If we only care about Chrome we could find a way to leverage the devtools protocol to query Power.Mac.BatteryDischarge.

If we want to compare to Safari it would be best to write a quick objc/swift program to do it. Optimally it would use something like Grand Central Dispatch to setup the sample collection to minimize the impact on the system.

The timer getting aggressively slacked would explain why measuring idle returns larger values than zero window safari!!

Durning the first 3% of discharge the number are quite unreliable. We think that's caused by OS policies to no have too many charge discharge cycles but not look like it's "not listening" to the user.

Needed are pandas and matplotlib