Switch to AUTORELEASE for simplicity.
Switch to building with Ninja for faster compilation.
Modify PKG_UNPACK instead of overriding Build/Prepare.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
We need to copy MagickWand-config, which handles pkg-config data, to the
staging dir, so that dependent packages can use it. php7-pecl-imagick,
for example, may otherwise use an eventual host-installed file, causing
compilation to fail. There's also a MagickCore-config file that does
the same thing for the MagickCore library, so a *-config wilcard is used
to handle all cases.
Since configure is called with --prefix=/usr, use $(SED) to change
definition of prefix and exec_prefix lines in *-config from /usr to
$(STAGING_DIR)/usr.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
The crude loop I wrote to come up with this changeset:
find -L package/feeds/packages/ -name patches | \
sed 's/patches$/refresh/' | sort | xargs make
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Since commit d18692c libxml2 (and by extension libxslt) is linked against
iconv. Now icecast needs access to iconv as well. Without it the build fails.
xslt.c: In function 'xsltSaveResultToString':
xslt.c:72:5: warning: this 'if' clause does not guard... [-Wmisleading-indentation]
if (result->children == NULL)
^~
xslt.c:75:2: note: ...this statement, but the latter is misleadingly indented as if it were guarded by the 'if'
buf = xmlAllocOutputBuffer(NULL);
^~~
xslt.c:81:27: error: dereferencing pointer to incomplete type 'xmlBuf' {aka 'struct _xmlBuf'}
*doc_txt_len = buf->conv->use;
^~
make[6]: *** [Makefile:545: xslt.o] Error 1
make[6]: Leaving directory '/openwrt/build_dir/target-mips_24kc_musl/icecast-2.4.4/src'
Signed-off-by: Sebastian Kemper <sebastian_ml@gmx.net>
Since commit d18692c libxml2 is linked against iconv. Now ices needs
access to iconv as well. Without it the build fails.
checking for ftime... yes
checking for XML configuration
checking for xml2-config... /builder/shared-workdir/build/sdk/staging_dir/target-aarch64_generic_musl/host/bin/xml2-config
checking for xmlParseFile... no
configure: error: Unable to link with libxml
Signed-off-by: Sebastian Kemper <sebastian_ml@gmx.net>
Fixes two CVEs relating to UPnP.
Removed libuuid dependency. It is not used.
Remove clock_gettime hack. It seems to have been fixed.
Removed upstream patches.
Refreshed the other ones.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
gst1-plugins-base might be required only for its libraries, not modules.
However, InstallDev tries to copy them unconditionally, failing when
no modules are selected/compiled.
Fixes#13973
Signed-off-by: Luiz Angelo Daros de Luca <luizluca@gmail.com>
These options are normally set to auto. Make them explicit to try to
fix buildbot errors.
Also make several options conditional on plugins being selected.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
These plugins were disabled in commit e35d46b. Add them back. The
required libs are added, too.
Closes#13545
Signed-off-by: Sebastian Kemper <sebastian_ml@gmx.net>
Reorganized Makefile a little bit for consistency.
Added PKG_BUILD_PARALLEL for faster compilation.
Refreshed patches.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Removed upstream ffmpeg backport.
Updated npupnp patch with latest upstream version.
Enabled CURL since it's needed for libnpupnp anyway.
Fixed up GCC filesystem CMake variables to be more accurate.
Removed npupnp include hack since it was properly fixed now.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Commit cb058bf changed gstreamer1's build options, and this broke some
applications.
First, -Doption-parsing=false is intended for very specific embedded
applications rather general packages. Please see:
https://github.com/openwrt/packages/pull/8847https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/388
Second, -Dregistry=false seems to break plugin discovery. Symptoms of
this include broken applications and gst-inspect-1.0 listing zero plugins.
Fixes#13180.
Signed-off-by: W. Michael Petullo <mike@flyn.org>
Diffrent file use its own setjmp_buffer, thus
we have two global variables called setjmp_buffer
I am not sure if we should use only one instance of it.
The patch sent upstream uses a similar approach.
https://sourceforge.net/p/minidlna/bugs/327/
Signed-off-by: Syrone Wong <wong.syrone@gmail.com>
Requires BUILD_PATENTED.
Added patch to fix compilation with it.
Removed include file hack. It's incorrect and it's also npupnp that
needs fixing.
Refreshed npupnp patch to the newest version sent upstream.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Added gettext-full host dependency. Normally it comes wth glib2, but in
the update to 2.65, it goes away. Add it here.
Refresh patch.
Add one for systems lacking distutils.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
This commit updates the backport from grilo-plugins master to include
a symbol name change.
Changes to libdmapsharing modified IS_X_... symbols to read X_IS_....
This was adopted in grilo-plugins master, as there was no official release
of grilo-plugins since adding support for libdmapsharing4 in the first
place. The next release of grilo-plugins will be the first to include
libdmapsharing4 support to include the X_IS_... name change.
Fixes issue mentioned at #12766.
Signed-off-by: W. Michael Petullo <mike@flyn.org>
Switch to meson as autotools support was removed.
Adjustment for consistency between packages.
Reorder plugins based on meson_options.txt order.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Updated options.
Removed deinterlace plugin as it's failing to build.
Reorganized plugin list based on meson_options.txt order.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
This is on track to replace libupnp in the next version. Since libupnp
is not building with GCC10, do this now.
Also backported needed header patch.
Replace other patches with the upstream versions.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Restore service triggers which got wrongly removed in commit
733aae9584 ("fix issues").
Without triggers, changing settings from LuCI or calling reload_config
from the cli won't have any effect.
Also adjust the uci commit call to only commit the minidlna configuration
when setting a UUID, to avoid committing unrelated user changes in other
config files.
Ref: https://github.com/openwrt/luci/issues/4194
Fixes: 733aae958 ("minidlna: fix issues")
Fixes: 37367bdc8 ("minidlna: create UUID in config if it is empty")
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Currently motion is failing to compile on bcm27xx because of a missing
dependency.
Sorted the configure options based on the order they appear in the
configure script. Also removed outdated ones.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Remove upstreamed patch.
Added patch to remove man and po stuff to fix compilation and speed up
the build.
Added PKG_BUILD_PARALLEL for faster builds.
Added PKG_INSTALL for consistency between packages. Used those install
paths as well to make it easier.
Other small cleanups for consistency between packages.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Only gerbera needs to read the file.
Added config file to conffiles to fix opkg update support.
Added several extra options to the UCI config. Manual editing is still required.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
For speexdsp, support for NEON on aarch64 was added in 1.2.0[1].
[1]: https://github.com/xiph/speexdsp/pull/8
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
With openwrt/openwrt@8dcc108760, the ARM
FPU compiler options are no longer part of CONFIG_TARGET_OPTIMIZATION.
This updates various packages that look for NEON/VFP support to search
CONFIG_CPU_TYPE instead.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
With the previous changes, my local build gets smaller:
From
12124957
To
12059421
Not that much but it's something.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
This is a fairly massive update that adds new dependencies.
Switched the CMAKE_OPTIONS to use ON/OFF as is done elsewhere.
Added a patch fixing compilation with libiconv-stub/full.
Add a small debug option to write to syslog debug information.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Presently, this attempts to disable atrac3, h264, hevc and vc1. These
encoders don't exist or are done through other programs (eg libx264).
Handling of those other programs is done elsewhere.
Signed-off-by: Ian Leonard <antonlacon@gmail.com>
This removes Python-related build variants, and adds
PYTHON3_PKG_BUILD:=0 and minor build adjustments (where appropriate),
for non-Python packages. There should be no changes to build output.
This also updates some include paths for python3-package.mk and/or
python3-host.mk to be relative to the package Makefile.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
Switched to meson.
Fixed license information.
Several Makefile cleanups for consistency between packages.
Enabled and added packages for a bunch of plugins.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Switch to meson.
No more static libraries as meson makes you choose.
Enabled speex and shout2 as the libraries are available.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Switched to meson.
Fixed license information.
No more static libraries as meson makes you choose.
Added several CONFIG_DEPENDS entries.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Switched to meson.
Fixed license information.
Rearranged some stuff for consistency between packages.
Removed static libraries. It seems meson only supports either one or the
other.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Provides access to services requiring a secure connection.
Chosen over openssl for license reason (can't redistribute). To my
knowledge, libressl will have the same issue.
mbed-tls is the only other ssl/tls provider, and this package is
complicated enough.
Signed-off-by: Ian Leonard <antonlacon@gmail.com>
Upstream thought it was a good idea to force downloads of grilo. Disable
it since it is not needed.
Removed two unnecessary variables.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Since getting rid of the patented line, it seems fdk-aac is now getting
enabled for -full, causing a missing dependency. OTOH, fdk-aac can only be
enabled if x264 is not. So add a ! dependency for it.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Both libraries in the tree (fdk-aac and faad) have patented functionality
disabled when CONFIG_BUILD_PATENTED is off.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Removed cover resizing patch. It doesn't work right.
Removed SIGHUP reload. While minidlna handles it, it needs work to
function properly.
Removed service_triggers. They don't work for similar reasons as above.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Removed inactive maintainer.
Cleaned up Makefile to remove old options.
Switched to PKG_INSTALL for consistency.
Added PKG_BUILD_PARALLEL for faster compilation.
Fixed license info.
Removed '' from enabled for consistency.
Ran init script through shellcheck. Batched config file writes.
Switched it to use procd. The -S parameter changes it to foreground. It
stands for systemd.
Added a sysctl tweak to get rid of warning.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
motion was previously forcing the use of libmicrohttp-ssl which is not
mandatory and it works just fine without SSL.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
better solution than using a static UUID by default
keep the default uuid empty then generate and save
a unique UUID on first start of minidlna service.
Signed-off-by: Alberto Bursi <bobafetthotmail@gmail.com>
minidlna uses UUID to uniquely identify the media server
by default, if the "uuid" option is not specified in
its setting file it will try to autogenerate it by
reading MAC address.
This seems to fail on some devices resulting in random
UUID on reboot, and this causes confusion and issues as
for clients this is a new server and must be added to
the list.
In OpenWrt there are also devices where the system can't
read the true MAC address and it is therefore randomized
on reboot.
So, add a static UUID in the settings file.
Since each mindlna server should have a different UUID,
if the user has more than one OpenWrt device with
minidlna, he should change the UUID of the
additional devices.
Signed-off-by: Alberto Bursi <bobafetthotmail@gmail.com>
Disable the init script by default to avoid log pollution; motion is
very verbose when it cannot open the configured camera.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Switched to codeload for simplicity.
Added patch to fix compilation without deprecated OpenSSL APIs.
Added PKG_BUILD_PARALLEL for faster compilation.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
This updates all Python packages that download their source from PyPi to
use pypi.mk.
This will allow future improvements/changes to pypi.mk to affect all
relevant packages.
This also makes it easier for future Python packages to start using
pypi.mk, when it's clear how it is used in existing packages.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
It has several compilation bugs in addition to being under no development.
There are also alternatives in the tree.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
The patch created patch files in the ffmpeg build directory, which did
absolutely nothing. Properly backported them.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
