Package: gnuzilla;
Reported by: Maxim Cournoyer <maxim <at> guixotic.coop>
Date: Sun, 31 Aug 2025 14:13:02 UTC
Severity: normal
Tags: patch
Message #17 received at 79355 <at> debbugs.gnu.org (full text, mbox):
From: Maxim Cournoyer <maxim <at> guixotic.coop> To: Mark H Weaver <mhw <at> netris.org> Cc: 79355 <at> debbugs.gnu.org Subject: Re: bug#79355: [PATCH v2] Update to 140.2.0-1. Date: Wed, 03 Sep 2025 23:14:42 +0900
Hey Mark!
Thanks for your reactivity!
Mark H Weaver <mhw <at> netris.org> writes:
[...]
>> * makeicecat: Update FFMINOR and SOURCEBALL_CHECKSUM.
>> Update patching for moved files on no longer existing patterns.
>> Do not set MOZ_APP_VENDOR/MOZ_APP_UA_NAME/MOZ_APP_PROFILE, as confvars, which now
>> fails the build. Delete no obsolete variables, or those already set to
>> the correct value in browser/moz.configure.
>> Do not patch (and break) translation data.
>> * data/patches/hide-firefox-view-button.patch: Update patch.
>> * data/patches/hide-submit-feedback-menu-item.patch: Likewise.
>> * data/patches/icecat-about-dialog.patch: Likewise.
>> * data/patches/moz-configure-changes.patch: Likewise. Set
>> MOZILLA_OFFICIAL to False by default, which disables telemetry.
>
> Interesting observation about telemetry! Can you please cite a
> reference for this claim, or else tell us how you came to know it?
I spent quite some time grepping the firefox source, and I randomly
found that odd relationship. In Telemetry.cpp, there's:
--8<---------------cut here---------------start------------->8---
// This function was copied from the implementation of
// nsITelemetry::isOfficialTelemetry, currently found in the file
// toolkit/components/telemetry/core/Telemetry.cpp.
static bool IsOfficialTelemetry() {
#if defined(MOZILLA_OFFICIAL) && defined(MOZ_TELEMETRY_REPORTING) && \
!defined(DEBUG)
return true;
#else
return false;
#endif
}
[...]
// Do not send the ping if we are not an official telemetry-enabled build;
// don't even generate the ping in fact, because if we write the file out
// then some other build might find it later and decide to submit it.
if (!IsOfficialTelemetry() || IsTelemetryDisabled()) {
return MaybeCache(cache, notificationType, notificationShown,
notificationAction, prevNotificationAction);
}
--8<---------------cut here---------------end--------------->8---
[...]
>> +--- a/browser/base/content/aboutDialog.css 2025-08-26 21:26:43.618669160 +0900
>> ++++ b/browser/base/content/aboutDialog.css 2025-08-26 21:31:54.207067616 +0900
>> +@@ -32,10 +32,9 @@
>> #rightBox {
>> background-image: url("chrome://branding/content/about-wordmark.svg");
>> background-repeat: no-repeat;
>> @@ -19,6 +19,9 @@
>> padding-top: 38px;
>> - margin-top: 20px;
>> + margin-top: 60px;
>> - }
>> -
>> + /* We don't want this box to contribute arbitrarily to the intrinsic size of
>> + * the dialog, so set the width to a reasonable size, but let it flex to take
>> + * all available space. */
>> +--- a/browser/base/content/aboutDialog.css
>> ++++ b/browser/base/content/aboutDialog.css
>> #rightBox:-moz-locale-dir(rtl) {
>
> The three lines above, which are the last three lines of
> 'data/patches/icecat-about-dialog.patch' after your patch is applied,
> are vestigial and should be dropped.
These lines are part of the context (from recent commit 0da0a471026ca),
so I'm not sure what is the suggestion here.
[...]
> It's a minor point, but I'd prefer to name this patch
> "fix-install.patch" without the "icecat-" prefix.
Done.
>> --- a/browser/moz.configure
>> +++ b/browser/moz.configure
>> -@@ -5,11 +5,11 @@
>> +@@ -5,16 +5,17 @@
>> # file, You can obtain one at http://mozilla.org/MPL/2.0/.
>>
>> imply_option("MOZ_PLACES", True)
>> -imply_option("MOZ_SERVICES_HEALTHREPORT", True)
>> -+imply_option("MOZ_SERVICES_HEALTHREPORT", False)
>> - imply_option("MOZ_SERVICES_SYNC", True)
>> +-imply_option("MOZ_SERVICES_SYNC", True)
>> -imply_option("MOZ_DEDICATED_PROFILES", True)
>> -imply_option("MOZ_BLOCK_PROFILE_DOWNGRADE", True)
>> -imply_option("MOZ_NORMANDY", True)
>> ++imply_option("MOZ_SERVICES_HEALTHREPORT", False)
>> ++imply_option("MOZ_SERVICES_SYNC", False)
>> +imply_option("MOZ_DEDICATED_PROFILES", False)
>> +imply_option("MOZ_BLOCK_PROFILE_DOWNGRADE", False)
>> +imply_option("MOZ_NORMANDY", False)
>> + imply_option("MOZ_PROFILE_MIGRATOR", True)
>> +
>> +
>> +-imply_option("MOZ_APP_VENDOR", "Mozilla")
>> ++imply_option("MOZ_APP_VENDOR", "GNU")
>> + imply_option("MOZ_APP_ID", "{ec8030f7-c20a-464f-9b0e-13a3a9e97384}")
>> ++imply_option("MOZ_APP_PROFILE", "mozilla/icecat")
>> + # Include the DevTools client, not just the server (which is the default)
>> + imply_option("MOZ_DEVTOOLS", "all")
>> + imply_option("BROWSER_CHROME_URL", "chrome://browser/content/browser.xhtml")
>
> I suggest adding the line:
>
> imply_option("MOZ_APP_UA_NAME", "Firefox")
>
> to 'browser/moz.configure' above, to compensate for the fact that the
> MOZ_APP_UA_NAME option had to be removed from 'browser/confvars.sh' (via
> the code in 'makeicecat').
This doesn't seem to be necessary; testing 140.2 my useragent is
currently:
--8<---------------cut here---------------start------------->8---
Mozilla/5.0 (X11; Linux x86_64; rv:140.0) Gecko/20100101 Firefox/140.0
--8<---------------cut here---------------end--------------->8---
According to code in nsHttpHandler.cpp, a fallback is used when it's not
set:
--8<---------------cut here---------------start------------->8---
mAppName.AssignLiteral(MOZ_APP_UA_NAME);
if (mAppName.Length() == 0 && appInfo) {
// Try to get the UA name from appInfo, falling back to the name
appInfo->GetUAName(mAppName);
if (mAppName.Length() == 0) {
appInfo->GetName(mAppName);
}
appInfo->GetVersion(mAppVersion);
mAppName.StripChars(R"( ()<>@,;:\"/[]?={})");
} else {
mAppVersion.AssignLiteral(MOZ_APP_UA_VERSION);
}
--8<---------------cut here---------------end--------------->8---
>
>> diff --git a/makeicecat b/makeicecat
>> index fc0bb76..85ca5c6 100755
>> --- a/makeicecat
>> +++ b/makeicecat
>> @@ -21,12 +21,13 @@
>>
>>
>> # metadata
>> -readonly FFMAJOR=128
>> -readonly FFMINOR=14
>> +readonly FFMAJOR=140
>> +readonly FFMINOR=2
>> readonly FFSUB=0
>> readonly FFBUILD=1
>> readonly GNUVERSION=1
>> -readonly SOURCEBALL_CHECKSUM='93b9ef6229f41cb22ff109b95bbf61a78395a0fe4b870192eeca22947cb09a53'
>> +# SHA-256
>
> I'd prefer to drop the "# SHA-256" comment. It's trivial enough to find
> this out by grepping for SOURCEBALL_CHECKSUM in the script.
The purpose of the comment was to tell the reader *which* checksum type
is used here. Maybe be obvious for some, but I had to check.
>> +readonly SOURCEBALL_CHECKSUM='956dce675c3b706d563caf07ed3ca9af632ab830be710dfd4351da78a0a2ef55'
>> readonly SOURCEBALL_SIGNINGKEY='14F26682D0916CDD81E37B6D61B7B526D98F0353'
>> readonly FFVERSION=${FFMAJOR}.${FFMINOR}.${FFSUB}
>> readonly SOURCEBALL=firefox-${FFVERSION}esr.source.tar.xz
>> @@ -41,7 +42,7 @@ readonly LIST_URL=lists.gnu.org/mailman/listinfo/bug-gnuzilla
>>
>> # l10n
>> readonly L10N_URL=https://github.com/mozilla-l10n/firefox-l10n
>> -readonly L10N_REV=fcd0300e8478d1ec4d1c097a073ddb8e1e0351e3
>> +readonly L10N_REV=4846304555affc92db4c693149c0f1c59afc7828
>
> I prefer to set L10N_REV to the commit specified in
> 'browser/locales/l10n-changesets.json', which in the case of 140.2 is
> a0c1ffda4ed19c0399e4175610361179e09829de.
Good catch! Adjusted.
>> @@ -541,7 +541,6 @@ configure_mobile()
>>
>> cat << EOF >> mobile/android/confvars.sh
>> # IceCat settings
>> -MOZ_APP_VENDOR=GNU
>> MOZ_APP_VERSION=${FFVERSION}
>> MOZ_PAY=0
>> MOZ_SERVICES_HEALTHREPORT=0
>
> Did you intend to make this change to the Android configuration? Note
> that our Android support has fallen into disrepair, and is not currently
> functional.
Yeah, it's not used right now, but if someone tried, it fail, so it's
better to remove it.
>> @@ -606,15 +605,6 @@ apply_branding()
>>
>> # Custom legal about pages
>> find l10n -wholename '*/brand.ftl' | xargs /bin/sed
>> 's/^trademarkInfo = .*/trademarkInfo = The IceCat logo is Copyright
>> 2008-2015 Free Software Foundation, released under the terms of the
>> GNU Lesser General Public License, version 3 or any later version./'
>> -i
>> - local string
>> - for string in rights-intro-point-2 rights-intro-point-3
>> rights-intro-point-4 rights-intro-point-5 rights-intro-point-6
>> rights-webservices rights-safebrowsing
>> - do
>> - find -name aboutRights.ftl | xargs sed -i "s/^${string}.*/${string} = /"
>> - done
>> - for string in helpus community-2 community-exp
>> - do
>> - find -wholename '*/browser/aboutDialog.ftl' | xargs sed -i "s/^${string}.*/${string} = /"
>> - done
>> cp "${DATADIR}"/aboutRights.xhtml toolkit/content/aboutRights.xhtml
>> cp "${DATADIR}"/aboutRights.xhtml toolkit/content/aboutRights-unbranded.xhtml
>> sed -i 's|<a href="http://www\.mozilla\.org/">Mozilla
>> Project</a>|<a href="http://www.gnu.org/">GNU Project</a>|g'
>> browser/base/content/overrides/app-license.html
>
> Are you sure that the code above is no longer appropriate? I see files
> named 'aboutRights.ftl' in l10n that still contain bindings for
> 'rights-intro-point-*', and therefore at least the first loop above
> still seems to be relevant.
I think the above code above was only clearing the last 3 keys (helpus,
comMnity-2, community-exp) with empty values, which is no longer valid
and broke the fluent parser (localization tool)
These values left unpatched are thus:
--8<---------------cut here---------------start------------->8---
community-exp = <label data-l10n-name="community-exp-mozillaLink">{ -vendor-short-name }</label> is a <label data-l10n-name="community-exp-creditsLink">global community</label> working together to keep the Web open, public and accessible to all.
community-2 = { -brand-short-name } is designed by <label data-l10n-name="community-mozillaLink">{ -vendor-short-name }</label>, a <label data-l10n-name="community-creditsLink">global community</label> working together to keep the Web open, public and accessible to all.
helpus = Want to help? <label data-l10n-name="helpus-donateLink">Make a donation</label> or <label data-l10n-name="helpus-getInvolvedLink">get involved!</label>
--8<---------------cut here---------------end--------------->8---
Which doesn't look too critical to my eye.
>> @@ -713,27 +703,18 @@ s/OpenSource/Free Software/g;
>> # Ditto for browser/modules/moz.build
>> sort_inner_list_in_file "EXTRA_JS_MODULES += [" "]" browser/modules/moz.build
>>
>> + # Adjust the capitalization of IceCat to match the expected symbol
>> + # names in build-time generated source files like TelemetryEventEnums.h.
>> + sed 's/IceCatview/Icecatview/g' -i browser/components/icecatview/metrics.yaml
>> +
>
> Hmm. Do you know where the "Icecatview" is originating from? Is there
> some code that changes all but the first letter of that identifier to
> lowercase?
I think it comes from the scripts under
toolkit/components/telemetry/build_scripts/gen_*.py, but I can't find
where the capital casing code is anymore.
>
>> # Copy js settings
>> cat "${DATADIR}"/settings.js >> browser/app/profile/icecat.js
>>
>> cat << EOF >> browser/confvars.sh
>> # IceCat settings
>> MOZ_APP_BASENAME=IceCat
>> -MOZ_APP_UA_NAME=Firefox
>> -MOZ_APP_VENDOR=GNU
>> MOZ_APP_VERSION=${FFVERSION}
>> -MOZ_APP_PROFILE=mozilla/icecat
>> -MOZ_PAY=0
>> -MOZ_SERVICES_HEALTHREPORT=0
>> -MOZ_SERVICES_HEALTHREPORTER=0
>> -MOZ_SERVICES_FXACCOUNTS=0
>> -MOZ_SERVICES_METRICS=0
>> MOZ_DATA_REPORTING=0
>> -MOZ_NORMANDY=0
>> -MOZ_SERVICES_SYNC=0
>> -MOZ_DEVICES=0
>> -MOZ_ANDROID_GOOGLE_PLAY_SERVICES=0
>> -MOZ_ADDON_SIGNING=0
>> EOF
>> }
>
> In the code above, what is the rationale for removing the settings for
> MOZ_PAY=0, MOZ_SERVICES_HEALTHREPORTER=0, MOZ_SERVICES_FXACCOUNTS=0,
> MOZ_SERVICES_METRICS=0, MOZ_DEVICES=0,
> MOZ_ANDROID_GOOGLE_PLAY_SERVICES=0, and MOZ_ADDON_SIGNING=0?
None of these outside MOZ_SERVICES_HEALTHREPORTER are still in use,
according to 'git grep' on Firefox's source.
++imply_option("MOZ_SERVICES_HEALTHREPORT", False) is now in the
'moz-configure-changes.patch'
There are two things that are currently broken in IceCat 140.2 based on
limited testing:
1. Contextual menu on right click has empty lines (sounds deja-vu? what was the past
resolution?)
2. The parameters page is borked. I have no idea what needs fixing.
I'd appreciate your help for the above two points.
--
Thanks,
Maxim
GNU bug tracking system
Copyright (C) 1999 Darren O. Benham,
1997,2003 nCipher Corporation Ltd,
1994-97 Ian Jackson.