Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
|
|
|
|
|
|
Otherwise, when PGO is ON, the build fails after './mach clobber'. It
looks like the build system does cleaning differently when .git dir is
present.
|
|
It looks like not all JavaScript that is available on the regular
DuckDuckGo page is free software. They have many projects in their
repos, but some of the JS from the search page has no clear licensing
info. So replaced the URL of the default search engine to HTML version
that currently doesn't have any JS at all.
|
|
Upstream Firefox now requires about-logo.svg in branding dir, so added
it by copying the same source file that is used for other logos.
LDFLAGS for i686 were changed the same way Arch Linux ARM did it.
Otherwise, the build fails due to this flag.
|
|
|
|
Build fails without them. Haven't found any binary blobs in these paths.
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Because data reporting is now hopefully disabled via build config
options, reviewed some of the related preference overrides:
Removed the following preferences:
datareporting.healthreport.uploadEnabled
defined only if MOZ_SERVICES_HEALTHREPORT is defined,
see modules/libpref/init/all.js
datareporting.healthreport.about.reportUrl
Removed upstream, see:
https://bugzilla.mozilla.org/show_bug.cgi?id=1352497
Can be found only in some test files
datareporting.healthreport.documentServerURI
Probably have been removed too, documentServerURI can
now be found only in obsolete documentation and some tests
healthreport.uploadEnabled
cannot be found alone in source tree.
Only datareporting.healthreport.uploadEnabled can be found.
datareporting.policy.dataSubmissionEnabled
defined only if MOZ_DATA_REPORTING is defined,
see modules/libpref/init/all.js
datareporting.healthreport.service.enabled
could find only in some test files
datareporting.policy.firstRunTime
datareporting.policy.dataSubmissionPolicyVersion
could not find in source tree
Also checked the following preferences:
toolkit.telemetry*
still available in the code, so have not removed them
pref("toolkit.telemetry.server_owner", "User");
This is just a nitpicking probably, because upstream preference says
"Mozilla" instead. And that is not true, because of the following
override in vendor.js :)
pref("toolkit.telemetry.server", "127.0.0.1");
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
MOZ_SERVICES_HEALTHREPORT and MOZ_NORMANDY are two missing bits to
undefine MOZ_DATA_REPORTING, which is defined if any one of
healthreporting, crashreporting, telemetry reporting, or Normandy are
set. Crashreporting and telemetry reporting have been already disabled.
Some data reporting related preferences and code are enabled when
MOZ_DATA_REPORTING is defined, for example.
These "features" are for Mozilla users only, they don't make much sense
for Iceweasel users. Probably even harmful for them and Mozilla too,
because I don't think Mozilla is interested in data reporting spam from
non-Firefox users.
I'm not sure if these features can be classified as back doors or
spyware in GNU FSDG terms, but that's another thing to consider.
Also added a check that fails the build if some of those features are
not disabled in configuration.
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Apart from upstream updates to list.json, there is also a bit of
preparation for the next significant change of how search engines are
handled upstream. 78.0 enabled this new mechanism, but 78.0.1 disabled
it again because of [1]. For details see [2]. It looks like Mozilla
can now update the list of available and visible search engines
remotely. So removing all engines except Wikipedia and DuckDuckGo just
to avoid surprises in future.
[1] https://bugzilla.mozilla.org/show_bug.cgi?id=1649393
[2] https://bugzilla.mozilla.org/show_bug.cgi?id=1542235
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
This is to reduce maintenance burden. The tab is going to have new
content in each new release, and as can be seen with this release, some
news items can contain references or mention nonfree stuff. Instead of
patching each new release, just removing it for now. Those who want to
see the news can find them (and more) on the upstream release notes
webpage.
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
_x86_64_skip_pgo can still be flipped manually for faster builds if
anybody wants that. Let's just follow upstream by default.
|
|
|
|
|
|
While Firefox Lockwise itself is libre-licensed (apps, website, etc),
the references to Android and iOS versions lead to Google Play and Apple
app stores, all of which execute some JavaScript that is not clearly
licensed, so can be considered nonfree.
These references can be found on about:logins page as items in "..."
menu, and on about:protections page in promo cards.
Removed them for now. Could also change the URLs to open the Lockwise
website, but although libre licensed, it still has large links to
nonfree app stores and Firefox on the front page, so that wouldn't be a
huge improvement.
|
|
Apart from the update according to the upstream changes, also removed
the reference to Firefox Developer Edition on Google Play because:
1. It appears to have the same freedom issues as regular Firefox;
2. Google Play webpage executes some JavaScript that's not clearly
licensed, so can be considered nonfree.
|
|
|
|
On i686 I've the following:
$ sudo pacman -Suu
:: Starting full system upgrade...
:: Replace haskell-exceptions with community/ghc-libs?
[Y/n]
resolving dependencies...
looking for conflicting packages...
error: failed to prepare transaction (could not satisfy
dependencies)
:: installing poppler (0.88.0-1.0) breaks dependency 'poppler=0.87.0'
required by cups-filters
:: installing icu (67.1-1.0) breaks dependency 'icu<66'
required by icecat
:: installing icu (67.1-1.0) breaks dependency
'icu=65.1' required by icedove
:: installing icu (67.1-1.0) breaks dependency
'icu<66' required by iceweasel
:: installing icu (67.1-1.0) breaks dependency 'icu<66'
required by texlive-bin
:: installing poppler (0.88.0-1.0) breaks dependency
'poppler=0.87.0-1.1' required by texlive-bin
More packages, that I don't have installed are probably affected.
Once we rebuilt and pushed the new packages, i686 users will be
able to upgrade to the new icu and poppler.
The new icu and poppler packages now have Provides for the
various libraries they provides:
$ pacman -S -i icu
Provides : libicudata.so=67-32 libicui18n.so=67-32
libicuio.so=67-32 libicutest.so=67-32
libicutu.so=67-32 libicuuc.so=67-32
$ pacman -S -i poppler
Provides : libpoppler.so=99-32 libpoppler-cpp.so=0-32
The provides are already there for x86_64 and armv7h.
Now that all architectures have provides for these two
packages, we will be able to convert our PKGBUILDs to depend
on the libraries provided, to make the next transition
easier: Thanks to the provides we can now provide packages
like icu-parabola, icu-57, etc to provide the current icu library
version.
This means that with the next icu library version that
comes from Arch Linux, Arch Linux 32, or Arch Linux ARM,
each packages that is built by Parabola and that has been
converted to use provides will continue to work as it will
pull the icu-parabola package to continue satisfying the
dependency on old libraries.
This in turn will enable users to continue upgrade will not
be blocked anymore by icu or poppler issues, and it will
enable developers to have more time to fix build issues.
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
|
|
sqlite and icu are leftovers from the times when it was built
--with-system-* flags. Works without these deps now and corresponds to
Arch Linux PKGBUILD.
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
All of them download and execute JavaScript code for which I couldn't
find any proof of their free licensing.
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
|
|
|
|
|
|
|
|
|
|
Show Private Browsing instead
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Please note checksum for patch hasn't been updated yet since more
changes are needed to it.
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
This reverts commit 87e2c81901db00a49ddc3faa967fefd2932141e4.
|
|
Tab"
This reverts commit 6401dce006aa7c922d3b4e0a20b0d2dc248ceaf5.
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Please note checksum for patch hasn't been updated yet since more
changes are needed to it.
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
This is mostly related to a recent switch of about:addons from XUL to
HTML [1].
extensions.webservice.discoverURL has been removed upstream, see [2].
extensions.getAddons.search.url seems to be removed as well, but could
not find any proofs except that source code search doesn't find it
anymore. Couldn't see any effect from it either (see [4]).
That's why removed those two. Also changed the search engines URL to a
more appropriate one, and also disabled 'Recommendations' section the
same way Tor Browser Bundle does this [3].
[1] https://bugzilla.mozilla.org/show_bug.cgi?id=1558982
[2] https://bugzilla.mozilla.org/show_bug.cgi?id=1544011
[3] https://trac.torproject.org/projects/tor/ticket/22073
[4] http://kb.mozillazine.org/Extensions.getAddons.search.url
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
i686 builds with gcc not clang, no need to remove debug flags for it.
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Single previous pref is not enough, effect is seen only with both.
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|
|
Signed-off-by: Andreas Grapentin <andreas@grapentin.org>
|