Don't unnecessarily gstreamer_system_package() components of gst-plugins-base that
are already linked to as system packages.
Do a single gstreamer_system_library() call as a micro-optimization, as it supports
multiple arguments for less internal calls to gstreamer_get_plugin_dir().
Closes: https://bugs.gentoo.org/834878
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
Don't unnecessarily gstreamer_system_package() components of gst-plugins-base that
are already linked to as system packages.
Do a single gstreamer_system_library() call as a micro-optimization, as it supports
multiple arguments for less internal calls to gstreamer_get_plugin_dir().
Closes: https://bugs.gentoo.org/866848
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
Drop wrong IUSE="introspection orc".
Remove gst-plugins-base dep that missed SLOT - the eclass depending on gst-plugins-bad
pulls in what's needed indirectly (toplevel meson.build hard-requires gst-plugins-base
libraries).
Closes: https://bugs.gentoo.org/805515
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
This is mainly for correctness, as the bad split builds didn't actually end up
building anything about shm or waylandsink as the subdirs of those are ignored
by the split packages builds.
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
To avoid enabling of waylandsink plugin build for all gst-plugins-bad splits
for correctness, we'll be adding it to eclasses extra plugin list that isn't
automatically parsed out of meson_options.txt. However gst-plugins-vaapi
also has an option named "wayland", which is about enabling wayland support.
Use the mechanism for making sure the eclass change won't force disable it
for gst-plugins-vaapi, even with USE=wayland.
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
With EAPI-8, python-any-r1 is handled by eclass, also adding py3.12 support.
Prepare for addition of wayland into gstreamer-meson.eclass extra options list.
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
gst-plugins-vpx, gst-plugins-voamrwbenc, gst-plugins-amrnb and
gst-plugins-x264 have preset files that we weren't installing.
This change makes them do so when they bump to EAPI-8.
Bug: https://bugs.gentoo.org/915220
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
I can't find any orc usage in the core module in recent history (including
autotools era 1.14), yet it even had pax-markings for it for tools, so I
assume it did use it long time ago?
It isn't used anymore and we actually get a QA warning about it after the
eclass fixups for false positives. While that QA warning still doesn't
cover it well for non-splits, it turns out it's actually right - it is not
used anywhere in the core module, so drop the USE flag and cleanup the
phases that aren't needed anymore after removing the orc conditional
pax-mark stuff.
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
gst-tester-1.0 is a similar case, but we don't have an easy way to disable build
on non-native ABI for that.
No other effect than skipping a couple compilation units on non-native ABI due
to how the installs are ordered by multilib eclasses (ending up with the native
version always on the filesystem before as well).
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
gst-plugins-good fails flvmux or other tests sometimes for me on timeout
otherwise, probably from running with a high amount of parallel tests
having some longer-running take even longer.
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
Upstream considers gles2 the better option and switches the default from opengl
to gles2 (when both are enabled) in 1.24. As it works better with GL drivers
in practive in 1.22 too, at least switch the default enabled over to it.
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
Upstream considers gles2 the better option and switches the default from opengl
to gles2 (when both are enabled) in 1.24. As it works better with GL drivers
in practive in 1.22 too, at least switch the default enabled over to it.
Signed-off-by: Mart Raudsepp <leio@gentoo.org>
Mart Raudsepp