Currently we set the following preferences: S1S2Test: telemetry_prompt = 999 if self.build.id < '20130103': telemetry_prompt = 2 prefs = { 'browser.firstrun.show.localepicker': False, 'browser.sessionstore.resume_from_crash': False, 'dom.ipc.plugins.flash.subprocess.crashreporter.enabled': False, 'browser.firstrun.show.uidiscovery': False, 'shell.checkDefaultClient': False, 'browser.warnOnQuit': False, 'browser.EULA.override': True, 'toolkit.telemetry.prompted': telemetry_prompt, 'toolkit.telemetry.notifiedOptOut': telemetry_prompt, 'datareporting.healthreport.service.enabled': False, } Smoketest: prefs = { 'browser.firstrun.show.localepicker': False, 'browser.sessionstore.resume_from_crash': False, 'browser.firstrun.show.uidiscovery': False, 'shell.checkDefaultClient': False, 'browser.warnOnQuit': False, 'browser.EULA.override': True, 'toolkit.telemetry.prompted': 999, 'toolkit.telemetry.notifiedOptOut': 999 } lacks 'datareporting.healthreport.service.enabled': False, Webappstartup: We don't set any preferences and only use the defaults since we don't have a way to create and specify a profile for the webapp. We should definitely set "app.update.enabled", False, "app.update.auto", False, What other prefs should we set? "browser.safebrowsing.enabled": False, "browser.safebrowsing.malware.enabled": False, "extensions.blocklist.enabled": False, What can we do about the webappstartup test?

I ran a local test with s1s2test and my nexus ones and gs3. You can see the results at: http://phonedash-dev.allizom.org/#/org.mozilla.fennec/throbberstop/local-blank/norejected/2015-03-31/2015-04-07/cached/errorbars/standarderror/notry Noticeable is that the throbber stop is lower especially for the twitter page but even to a lesser degree the blank page. It does appear the results are less noisy but I won't guarantee that this will translate to the production devices. I found most of these preferences by digging through the reftest preferences in https://dxr.mozilla.org/mozilla-central/source/layout/tools/reftest/reftest-preferences.js Is the decreased throbber stop time something to be expected from these preference changes? snorp, mfinkle: Would you look and see if these are ok for you? They will only apply to the s1s2test at the moment. I still need to figure out the webapp profile situation.

fyi, pref0 is with the current preferences while pref1 is with the new preferences.

That's certainly worth noting when preference changes yield "improvements" so I'm glad you pointed this out. Looking at the list, I see a few prefs that could affect pageload: "image.high_quality_downscaling.enabled" - especially for twitter "image.single-color-optimization.enabled" - especially for twitter "layout.interruptible-reflow.enabled" I also se there is a pref to control the cursor blink time ("ui.caretBlinkTime") which might be useful for Fennec to control power usage too.

As to whether they are OK to use or not, I'm not sure. If we want to measure performance of what we ship, we may not want to turn those prefs on. More thought required.

mfinkle, are you looking at the full list in the reftest-preferences.js file? I used a limited set which I picked based on whether they involved external lookups for updates etc. See https://github.com/bclary/autophone/commit/a2a078f6f11117c49ca46ce85d29cc1deb7844de for the exact commit.

(In reply to Bob Clary [:bc:] from comment #5) > mfinkle, are you looking at the full list in the reftest-preferences.js file? I was looking at the full list. > > I used a limited set which I picked based on whether they involved external > lookups for updates etc. See Some of those look unused in native Fennec: browser.firstrun.show.localepicker browser.firstrun.show.uidiscovery Some are Desktop specific: browser.newtabpage.directory.ping browser.newtabpage.directory.source These could affect startup + initial pageload: browser.safebrowsing.enabled browser.snippets.enabled

mcote: I updated the branch/pr removing browser.firstrun.show.localepicker rowser.firstrun.show.uidiscovery browser.newtabpage.directory.ping browser.newtabpage.directory.source We need a better approach to preferences, but I'd like to defer that to a different bug.

If I understand correctly, mochitests override a much larger set of prefs, via testing/profiles/prefs_general.js. I do not know if there is value in looking at that list, but wonder, were those prefs considered?

gbrown: No, I hadn't considered them. Let me look now.

mfinkle: Would you mind doing a quick scan of the attached document. It lists the preferences included in Autophone to that in the testing prefs_general.js file.

I'll also need pref("xpinstall.signatures.required", false);

https://github.com/mozilla/autophone/commit/fd2c00008f66ad42f52b5eb8251d38454c678538 with added xpinstall.signatures.required pref. -> fixed. We can follow up with additional prefs when I refactor the prefs as part of the webappstartup profile bug 1152392