some additional settings

[crssi]

Just putting here some settings I have stumbled upon.
I haven't investigate into those yet, so don't kill me, please. :)
Some might be deprecated already, some might not be existing, etc.

• browser.addon-watch.deactivate-after-idle-ms
  browser.addon-watch.interval
  browser.addon-watch.percentage-limit
  • wontfix: commented by @earthlng: only for Nightly; we don't need this
• browser.customizemode.tip0.shown
  • wontfix: I would guess, like a firstrun, its already been triggered for everyone
• browser.formfill.expire_days
  • see ToDo: compare vs pyllyukko user.js, decide on additions #10
• browser.sessionstore.enabled
  • wontfix: deprecated
• browser.xul.error_pages.enabled
  • wontfix: commented by @earthlng: false is bad, true is default; we don't need this
• network.http.use-cache
  • wontfix: deprecated
• toolkit.telemetry.optoutSample
  • wontfix: deprecated and not messing with telemetry settings (just master switches, urls, ids)
• toolkit.telemetry.previousBuildID
  • wontfix: deprecated
• toolkit.telemetry.rejected
  • wontfix: deprecated
• toolkit.telemetry.server_owner
  • wontfix: not messing with telemetry settings (just master switches, urls, ids)
• browser.download.manager.quitBehavior
  • wontfix: see http://kb.mozillazine.org/Browser.download.manager.quitBehavior
    don't really know what would be the best for wider audience... or default (the value = 0) or is this privacy issue (then the value = 2)
• media.hardware-video-decoding.enabled
  • wontfix: not messing with video playback (only those accessible via options)
• dom.maxHardwareConcurrency
  • wontfix: at this stage

==

• browser.sessionstore.restore_on_demand
  • NFI .. why do we get lists dumped on us

I am out for a few days.
If anyone already know for some of those, I would be happy for your comment.

[earthlng]

browser.xul.error_pages.enabled is pretty terrible when set to false, and it's true by default.
We could enforce true and just add it with the other browser.xul.error_pages or not include it at all.

addon-watch is only enabled in Nightly and used for performance monitoring, I don't think we need this

[earthlng]

https://dxr.mozilla.org/mozilla-central/source/modules/libpref/init/all.js#5557

// return the maximum number of cores that navigator.hardwareCurrency returns
pref("dom.maxHardwareConcurrency", 16);

[Atavic]

pyllyukko/user.js#181

[earthlng]

heh .. it's [text] and then (url) .. not [url] then (the string "url") .. that just makes the hyperlink point back to this article .. lulz

what? I copied the URL (dxr) into the Write field and submitted. The link works for me or did you correct it in the meantime? If it's about the edited OP comment, I didn't edit that at all, I think crssi added the info himself based on my comment.

[earthlng]

browser.sessionstore.restore_on_demand default is true and IMO that's the preferred value.
afaik Session Manager also uses that pref and it's used to not reload restored tabs. Maybe this only makes sense when cache is enabled, idk.
@crssi only dropped us a list of prefs, idk if he suggests we set this to false. (??)

[earthlng]

dom.maxHardwareConcurrency I'm not sure it would impact performance at all. FF itself is in charge of how many cores it wants to use, and maybe this pref only exposes the number of cores to sites.
It's a max value, so it would only make sense for spoofing purposes if we set this very low.
If that's indeed the case, we would need a site with the most common core-numbers.
Idk, maybe this deserves some more digging, but I don't mind if we discard it.
The issue that @Atavic linked to, mixes number of cpus with number of threads. limiting the number of threads a serviceworker can use is of course impacting performance, but spoofing the amount of cores probably shouldn't.

[Atavic]

Demo test page: https://github.com/oftn-oswg/core-estimator

Core Estimator reads navigator.hardwareConcurrency with .google-analytics.com/ga.js
then opens up a progressive number (1-2-4-8...) of WebWorkers to give an extimation of the number of cores (by looking at the time needed to execute the webworkers).

This is a Javascript Performance Fingerprinting, as noted in the Draft The Design and Implementation of the Tor Browser .

• https://html.spec.whatwg.org/multipage/workers.html#navigator.hardwareconcurrency