1963545 - Replace nsContentUtils::SpoofLocaleEnglish with ShouldResistFingerprinting(RFPTarget::JSLocale)

Status: RESOLVED FIXED

(Core :: Privacy: Anti-Tracking, task)

Description

[Tom Schuster [:tschuster]]

ShouldResistFingerprinting(RFPTarget::JSLocale) already uses the pref privacy.spoof_english under the hood, so I don't think we need the older method.

Comment 1

[Tom Schuster [:tschuster]]

Attachment: Bug 1963545 - Replace nsContentUtils::SpoofLocaleEnglish with ShouldResistFingerprinting(RFPTarget::JSLocale). r?fkilic

Comment 2

[Tom Schuster [:tschuster]]

test_SVG_switch_language_spoof.html fails now. Are we maybe caching the RFP state?

Comment 3

[Fatih Kilic [:fklc]]

I don't think so. Do .html tests run with system principal? nsContentUtils::SpoofLocaleEnglish didn't used to check for principal, maybe aDoc->ShouldResistFingerprinting(RFPTarget::JSLocale) returns false because the doc is chrome?

Comment 4

[Tom Schuster [:tschuster]]

No, the test isn't running with system principal. The problem is that we actually change the preference intl.accept_languages to English when setting the privacy.spoof_english to 2. It would be nice to remove that code and instead change all users of intl.accept_languages to respect RFPTarget::JSLocale instead.

I worked around this by switching around the order of the test.

Comment 5

[Tom Schuster [:tschuster]]

This patch changes our behavior so that we only respect privacy.spoof_english = 2 when FPP or RFP is enabled. Previously privacy.spoof_english was documented to work in conjunction with RFP, but didn't actually require privacy.resistFingerprinting.

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/98df7963c3f1
Replace nsContentUtils::SpoofLocaleEnglish with ShouldResistFingerprinting(RFPTarget::JSLocale). r=fkilic

Comment 7

[Sandor Molnar[:smolnar]]

https://hg.mozilla.org/mozilla-central/rev/98df7963c3f1

Comment 8

[Pier Angelo Vendrame]

(In reply to Tom Schuster (MoCo) from comment #5)

This patch changes our behavior so that we only respect privacy.spoof_english = 2 when FPP or RFP is enabled. Previously privacy.spoof_english was documented to work in conjunction with RFP, but didn't actually require privacy.resistFingerprinting.

Sorry, probably a bit OT for this Bug, but why is test changing this pref a problem?
I got several test failures I still haven't managed to check because of this in the tests I wrote for my Bug 1666613 patch.

Comment 9

[Tom Schuster [:tschuster]]

Sorry, probably a bit OT for this Bug, but why is test changing this pref a problem?

Which pref, do you mean intl.accept_languages or the privacy prefs? For the former it's problematic because when a pref is changed in the background, without e.g. using SpecialPowers.pushPrefEnv, that might influence other tests. In this specific case I think it's also a bit confusing, because it doesn't even seem necessary at first glance.

Comment 10

[Pier Angelo Vendrame]

(In reply to Tom Schuster (MoCo) from comment #9)

Sorry, probably a bit OT for this Bug, but why is test changing this pref a problem?

Which pref, do you mean intl.accept_languages or the privacy prefs? For the former it's problematic because when a pref is changed in the background, without e.g. using SpecialPowers.pushPrefEnv, that might influence other tests. In this specific case I think it's also a bit confusing, because it doesn't even seem necessary at first glance.

Yes, I meant intl.accept_languages.
It makes sense, thanks.