Consider using a screen WakeLock for running speedometer

[julienw]

I noticed when running on mobile that the screen is quickly dimming and even possibly turned off during the course of the run. I was wondering if the mobile's performance could be automatically reduced when the screen is off.

Therefore I was thinking that using a screen WakeLock could be beneficial. One issue is that it's not supported in Firefox though I believe it could be implemented. But this shouldn't be a problem for comparisons if the user turns off the screen dimming/screen off. Although theorically keeping a screen WakeLock could also bump up some performance on mobile? I honestly have no idea.

I'd like to see what y'all think of this.

[bgrins]

Having wake lock would definitely make it easier to run on mobile, although I think it’s quite hard to run outside of CI regardless, due to the total runtime.

Is the ability for people to run this on mobile devices from the website important to this group? If so we should also consider other ways to make it less difficult (reducing total iterations, perhaps).

As far as changing the code to use wake lock, I think we can’t practically do that until

1. all engines support it (so that any performance changes / functionality is shared)
2. WebKit is able to use secure context APIs in CI (see also having to remove use of the crypto api for uuid generation)

[bgrins]

Another thing we’ve discussed in the past that may help is creating a runbook for reliably running and reporting on results. We can include instructions like disabling screen lock, having the phone plugged in, etc there.

[rniwa]

Seems reasonable.

[julienw]

This bug tracks the implementation of the WakeLock API in Firefox: https://bugzilla.mozilla.org/show_bug.cgi?id=1589554