1264796 - Exclude users with script timeout prefs set to 0 from e10s meta experiments

Status: RESOLVED WONTFIX

(Firefox :: General, defect)

Description

[Jim Mathies [:jimm]]

To get better data related to SLOW_SCRIPT_PAGE_COUNT rates, we should disqualify any user who's adjusted prefs such that the slow script dialog won't display.

http://mxr.mozilla.org/mozilla-central/source/modules/libpref/init/all.js#2588

prefs:
dom.max_script_run_time
dom.max_child_script_run_time

If either of these are 0, the user should be excluded.

Comment 1

[Jim Mathies [:jimm]]

Felipe, can we get something like this into the 47 experiment?

Comment 2

[:Felipe Gomes (needinfo for replies!)]

Sure thing

Comment 3

[Jan de Mooij [:jandem]]

What happens if the pref changes after the experiment starts, will they no longer be part of it? Will this happen immediately or at the next browser restart?

The non-e10s slow script dialog has a "Don't show this again" option, which sets the pref, so if the user is still part of the experiment after that point (even if it's until the next restart) the results are unfair.

Comment 4

[Brad Lassey [:blassey] (use needinfo?)]

I think this is the wrong approach, I don't see a reason to exclude them from the experiments as a whole, we should just be excluding their uptime from the measurement of slow script dialogs. Also... we should do this for pref setting other than 10, rather than just for 0.

Comment 5

[:Felipe Gomes (needinfo for replies!)]

Wontfixing in favor of collecting the pref (bug 1265000) and looking at it on data analysis time.