logcat spits out fully drawn times like, "+551ms". Sometimes, it does, "+551ms (total 1m331s)". For the same launch command, this behavior appears to change between devices: the G5 uses the latter and the P2 uses the former.
I didn't realize this so I landed https://github.com/mozilla-mobile/FNPRMS/pull/20 to switch from non-total times to total times to support the G5: instead, we should actively support both. We should take total if it's available, otherwise fall back to non-total. This code is in times.py and we should add this ability to both fennec-nightly and fennec code.
In the current implementation, our G5 metrics should work correctly but runs on other devices may not, including runs if people use FNPRMS for local testing. Applink is unaffected & should work in all configurations.
logcat spits out fully drawn times like, "+551ms". Sometimes, it does, "+551ms (total 1m331s)". For the same launch command, this behavior appears to change between devices: the G5 uses the latter and the P2 uses the former.
I didn't realize this so I landed https://github.com/mozilla-mobile/FNPRMS/pull/20 to switch from non-total times to total times to support the G5: instead, we should actively support both. We should take total if it's available, otherwise fall back to non-total. This code is in times.py and we should add this ability to both
fennec-nightly
andfennec
code.In the current implementation, our G5 metrics should work correctly but runs on other devices may not, including runs if people use FNPRMS for local testing. Applink is unaffected & should work in all configurations.