USE=modules was declared in IUSE directly instead of via
MODULES_OPTIONAL_IUSE. The linux-mod-r1 eclass only gates its exported
phases (pkg_setup, src_compile, src_install, pkg_postinst) through
_modules_check_function when MODULES_OPTIONAL_IUSE is set; without it all
kernel-related code runs unconditionally regardless of USE=-modules. This
causes fatal failures in environments without a configured kernel source
tree (e.g. catalyst livecd-stage1, prefix installs, chroots).
The eclass documentation notes: "The typical recommended value is
+modules" and "modules being optional can be useful even if user space
tools require them (e.g. installing in a chroot or prefix when the
modules are loaded on the host)."
Also fix the MODDULES_KERNEL_MIN typo (double O) introduced in the same
commit that added USE=modules; the eclass reads MODULES_KERNEL_MIN so the
minimum kernel version check of 6.16 was silently ignored.
Signed-off-by: Nic Boet <nic@boet.cc>
Part-of: https://github.com/gentoo/gentoo/pull/46389
Closes: https://github.com/gentoo/gentoo/pull/46389
Signed-off-by: Sam James <sam@gentoo.org>
This release fixes building w/ motif-2.5.2.
No support for multiple audio backends, except jack+alsa, and the patch
previously applied results in unused_functions. To no break profiles w/
`?? ( alsa pulseaudio)`, use alsa as a fallback w/ the disadvantage of
having alsa-lib required in any case.
install s7.h and export symbols to fix s7 w/o gui/notcurses
notcurses requires s7
remove readline, this is no longer an option
update x11-libs for gui
opengl was automagic, update deps (glu and libglvnd) and add an option
filter lto with opengl (type-mismatch)
jack requires libsamplerate
define the path for all tools avoiding automagic; if one is missing
during runtime, snd shows an explicit error in console (no crash)
noexecstack is no longer required
configure only checks <=ruby-3.0, add a sed-op to try w/ the latest
version installed
add a desktop file w/ an icon
install manpages
always install html
Closes: https://bugs.gentoo.org/878307
Closes: https://bugs.gentoo.org/879607
Closes: https://bugs.gentoo.org/887149
Closes: https://bugs.gentoo.org/914433
Closes: https://bugs.gentoo.org/965500
Closes: https://bugs.gentoo.org/976384
Signed-off-by: Nicolas PARLANT <nicolas.parlant@parhuet.fr>
Part-of: https://codeberg.org/gentoo/gentoo/pulls/1120
Merges: https://codeberg.org/gentoo/gentoo/pulls/1120
Signed-off-by: Sam James <sam@gentoo.org>
Let's welcome the new NIH one-fix fork of dev-python/nest-asyncio
that is now required by dev-python/ipykernel. Perhaps all the remaining
IPython packages will eventually migrate too, and we'll be able not to
mix two nest-asyncio versions. Well, at least they've renamed
the module, so we don't have to play transition games and miraculously
they did not bother changing (and breaking) the PEP 517 backend
in the process, which they would probably never fix.
Signed-off-by: Michał Górny <mgorny@gentoo.org>
sam confirmed that it does not actually link with it when bindist
is set.
Also disable the system library check so it doesn't try to use
it when missing, it shouldn't actually *use* bundled either when
bindist is set (even if cmake says it will).
Won't revbump over this due to the long build so, if want to
depclean it, please manually rebuild.
Thanks-to: Sam James <sam@gentoo.org>
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
This package doesn't actually use clang but it is used indirectly
through ffmpeg and I feel it is better to avoid confusion given
USE=cuda is typically associated with nivdia-cuda-toolkit instead.
Also simplifies, no need to have cuda? ( ffmpeg[cuda-clang] ).
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
USE=cuda has a lot of implications because it typically means pulling
nvidia-cuda-toolkit, which is notably masked in stable and causing
trouble/confusion for stable users here.
Rather than unmask, I feel we should rename it to use the same that
some other ebuilds been using.
May disrupt some users given we do not have a USEmove mechanism yet.
Hopefully users that had this stop working will think to look at USE
(may not be worth having a postinst message over, should have few
users and these often go overlooked anyway -- sorry for any issues
still).
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>