Looks like the (graft) update to gtk+ to 3.24.49 never built on i686-linux due
to a failing test. It is unclear why, but since we use an old librsvg on this
platform compared to x86_64, that is a major difference in the build/tests.
* gnu/packages/gtk.scm (gtk+)[arguments]<#:phases>: Skip the linear-gradient
on i686-linux.
Change-Id: I61e2fccaeaa92889c5519ad7845483f0e6a6715f
Printers managed by CUPS might require supplementary files to function,
such as color profiles or filters. CUPS checks permissions on such files
to prevent the execution of unsafe code. One of the conditions-that the
files are owned by root-must be short-circuited on Guix, because this
condition cannot be met on a system with an unprivileged daemon (where
store files are owned by `guix-daemon`).
* gnu/packages/patches/cups-relax-root-ownership-check.patch: New patch.
* gnu/local.mk (dist_patch_DATA): Register it.
* gnu/packages/cups.scm (cups)[source]: Include it.
Change-Id: I77f67f996d057a34bd018ab97cda54577060b0c3
Signed-off-by: John Kehayias <john@guixotic.coop>
* gnu/packages/xdisorg.scm (libxkbcommon): Update to 1.11.0.
[source]: Use git-fetch instead of url-fetch.
[native-inputs]: Add xorg-server-for-tests and xkbcomp. Reformat.
[arguments]<#:configure-flags>: Add flag to enable docs.
(libxkbcommon-1.5, libxkbcommon-1.8): Remove variables.
* gnu/packages/terminals.scm (foot)[inputs]: Use libxkbcommon instead of
libxkbcommon-1.8.
Change-Id: I0e4ce4cab04c059916ea7dddd31d2780c7b49b26
Fixes CVE-2025-58060 and CVE-2025-58364.
* gnu/packages/cups.scm(cups): Update to 2.4.14.
[source]: Drop unneeded patch.
* gnu/packages/patches/cups-minimal-Address-PPD-injection-issues.patch: Remove
it.
* gnu/local.mk (dist_patch_DATA): Deregister it.
Change-Id: I719e568716c8739aca16c6ebc29f50c7d2ac83bc
* gnu/packages/documentation.scm (doxygen): Update to 1.14.0.
[arguments]<#:phases>: Rename disable-bibtex-test to disable-some-tests and
disable a failing test (reported upstream).
Change-Id: I2402138cb20916601adbc88ba9a769520d673f7f
These tests fail for unknown reasons locally (even on master, without the
changes from this mesa-updates branch) though presumably pass on some machines
since there are substitutes available. It appears to be non-deterministic.
See discussion at <https://codeberg.org/guix/guix/issues/1257>.
* gnu/packages/autotools.scm (automake)[arguments]<#:phases>: Skip some
tests.
Change-Id: I87099386483efdbb62e42251298f0995f3967179
The following CVEs were fixed:
- CVE-2025-24855: Fix use-after-free of XPath context node
- CVE-2024-55549: Fix UAF related to excluded namespaces
Additionally, a patch from Debian for generated IDs and reproduciblity no
longer applies cleanly and is likely problematic or (partially?) unneeded.
See <https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=902051> for the latest.
* gnu/packages/xml.scm (libxslt): Update to 1.1.43.
(source): Remove patch.
* gnu/packages/patches/libxslt-generated-ids.patch: Delete it.
* gnu/local.mk (dist_patch_DATA): Unregister it.
Change-Id: Ia10d906bab090792d28524beda6aca79a5a21684
This test fails on btrfs more generally, not only on hurd.
* gnu/packages/base.scm (coreutils)[arguments]<#:phases>: Skip the
reflink-auto test unconditionally.
Change-Id: I27c4750f848a6b4c3c251614d95aefe41efe5d81
Python bindings for libxml2 are now built as part of libxml2; remove all
python-libxml2 inputs for libxml2 (nearly all already had libxml2) and
deprecate the package. This was done manually using
git grep python-libxml2 gnu/packages
to find affected packages and then check all their input fields.
With the update to libxml2, we can remove all the libxml2-next* variants.
This change was automated with:
git grep -rl libxml2-next | xargs sed -i 's/\(libxml2-next\|libxml2-next-for-grafting\)/libxml2/g'
* gnu/packages/xml.scm (libxml2): Update to 2.14.6.
[arguments]<#:phases>: Add configure-python phase (from python-libxml2).
[native-inputs]: Add pkg-config and python-minimal.
(libxml2-next, libxml2-next/fixed, libxml2-next-for-grafting): Remove.
(python-libxml2): Deprecate in favor of libxml2.
Change-Id: I2976f2fde086e9bc345d661edbdc125abbb7a753
* gnu/packages/gtk.scm (gtk+): Update to 3.24.43.
[replacement]: Remove.
[source]: Use git-fetch.
[arguments]<#:phases>: Add some test to disable-failing-tests phase.
(gtk+/fixed): Remove.
Change-Id: Ia34ac538f36424748e0a813cfc4e0dad2d643d0b
Upstream no longer uses autotools so we build with meon-build-system now. The
configure flag to not build a static output is no longer supported and even
though it was used previously, there was still a static library built.
Finally, the gnuplot configure flag is also removed, per
<https://issues.guix.gnu.org/69962>.
* gnu/packages/xdisorg.scm (pixman): Update to 0.46.4.
[build-system]: Switch to meson-build-system.
[arguments]<#:configure-flags>: Rewrite with gexp and for meson. Remove
"--disable-static" as it is not a supported flag. Remove "--enable-gnuplot"
flag.
[inputs, native-inputs]: Remove.
[home-page]: Use https URL.
Change-Id: I6bc17761ce96de9ef17bb8b420b8570d1ea11e07
Without turning off this error, the build fails on some Vulkan related code.
* gnu/packages/video.scm (ffmpeg)[arguments]<#:phases>: In
relax-gcc-14-strictness phase, add -Wno-error=int-conversion to CFLAGS.
Change-Id: I3b0d0f5cbc80aae7b4b164b9a17bd0d8804c9fdd
The check phase started failing (after core-updates merge maybe?) as there are
no tests since building them was previously disabled already.
* gnu/packages/vulkan.scm (shaderc)[arguments]: Disable tests.
Change-Id: Ibfa53bd30e90a3b4fd6b5bbed403b8a9aa615e9c
Note: this is not the first "stable" release, but making this available for
testing as some configuration we used have been dropped upstream.
* gnu/packages/gl.scm (mesa): Update to 25.2.0.
[arguments]<#:meson>: Use meson-1.8.
<#:configure-flags>: Remove osmesa and gallium-xa flags (no longer available).
* gnu/packages/rust-crates.scm (mesa-cargo-inputs): Add rustc-hash.
Change-Id: I36d59582b5d827f78040a248a54fd11e2f982393
