Update OpenCV recipe from version 3.2 to 3.3. Rebase patches and drop
those which are no longer relevant or already contained in upstream
OpenCV. Fix changed build options, such as UPDATE_PROTO_FILES to
PROTOBUF_UPDATE_FILES. Update also subcomponents (opencv-contrib and
opencv-3rdparty) to match with 3.3.
Signed-off-by: Ismo Puustinen <ismo.puustinen@intel.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Dont hardcode cpp to point to gcc collection, helps
compiling with clang
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Add missing Zlib to license list
Add patches to fix build issue with gcc7 and missing internal build dependency
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
License checksum changed due to opyright year changed to 2017
Fix build with clang/hardening
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* remove recipes which support only bluez4 and are blacklisted when
bluez5 is selected in DISTRO_FEATURES, if someone is still using
bluez4, then it can be restored in separate meta-bluez4 and
maintained by people using it.
* there are few recipes which support both bluez5 or bluez4 based on
selected DISTRO_FEATURES, these can stay in meta-oe repository, but
now people without bluez5 in DISTRO_FEATURES will need to add
meta-bluez4 layer to satisfy bluez4 dependency
meta-gnome/recipes-connectivity/obex/obexd_0.48.bb:DEPENDS += "${@bb.utils.contains('DISTRO_FEATURES','bluez5','bluez5','bluez4',d)}"
meta-gnome/recipes-gnome/gnome-bluetooth/gnome-bluetooth_3.18.2.bb: ${@bb.utils.contains('DISTRO_FEATURES','bluez5','bluez5','bluez4',d)} \
meta-oe/recipes-connectivity/obex/obex-data-server_0.4.6.bb:DEPENDS += "${@bb.utils.contains('DISTRO_FEATURES','bluez5','bluez5','bluez4',d)}"
meta-oe/recipes-connectivity/obex/openobex_1.7.2.bb:DEPENDS_append_class-target = " ${@bb.utils.contains('DISTRO_FEATURES','bluez5','bluez5','bluez4',d)}"
meta-oe/recipes-connectivity/packagegroups/packagegroup-tools-bluetooth.bb:RDEPENDS_bluez4 = " \
meta-oe/recipes-connectivity/packagegroups/packagegroup-tools-bluetooth.bb:# Install bluez4 tools or bluez5 tools depending on what is specified in the distro.
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
pkg_postinst_uim-gtk2.0 is not worked. Plus, uim-gtk3 is not supported.
oe-core has gtk-immodule-cache class for update the immodule.cache.
Modified to use the this class instead of pkg_post_uim-gtk2.0 and added support to uim-gtk3.
Signed-off-by: Yusuke Mitsuki <mickey.happygolucky@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
pkg_postinst*/pkg_prerm_* functions are not worked correctly.
Because uim-module-manager executes on the host machine.
It needs executed via qemu.
Signed-off-by: Yusuke Mitsuki <mickey.happygolucky@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
These functions are not worked correctly.
However default im is installed. thus we determined it not need.
Signed-off-by: Yusuke Mitsuki <mickey.happygolucky@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
To uim-anthy is working correctly needs register by uim-module-manager.
However 'uim-module-manager --register anthy' is failed.
Because uim-module-manager uses iconv(EUCJP to UTF8) internally.
iconv needs glibc-utils and EUCJP conversion needs glibc-gconv-euc-jp.
Signed-off-by: Yusuke Mitsuki <mickey.happygolucky@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
anthy.h could not found at confiugre of uim.
class-target override was used incorrectly with +=,
but instead of expected append to DEPENDS set above
it overrided whole DEPENDS for target builds so anthy
was missing.
Signed-off-by: Yusuke Mitsuki <mickey.happygolucky@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
The uim-anthy package is created that empty.
To avoid this problem by defining a packages with a rough PATH list after the uim-anthy package.
Signed-off-by: Yusuke Mitsuki <mickey.happygolucky@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
On some host (ubuntu 1404), build manual failed.
...
./tools/man-generator --primary lvscan lvscan.8_des
Failed to stat description file lvscan.8_des.
...
Do not build man to workaround the issue.
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Upstream added new dependency on udev in
a7ae55d NativeStateDRM: Probe the DRM node path to use with udev
Patch removed since it was merged upstream in
5fcdba1 wayland: Fix destruction order of surface-related objects
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
If this recipe was included in an eSDK and then eSDK incremental updates were
generated by repeatedly running the oe-publish-sdk script, after installing the
first of such updates, and any following as well, with 'devtool sdk-updates' the
eSDK gets corrupted. Indeed some of the files listed in SRC_URI (namely anything
in *.patch and *.service) get deleted by those updates, with bitbake that starts
warning with 'Unable to get checksum for android-tools* SRC_URI entry' at every
parse. Since the files are missing, such warnings turn into errors if trying to
bake android-tools.
The root cause is the somewhat unusual presence of android-tools/.gitignore,
indeed ignore everything not explicitly excepted. When oe-publish-sdk creates
the git repo it relies on to feed eSDK updates, the .gitignore gets honored,
leading to the buggy behaviour above.
This patch adds *.patch and *.service to the excepted patterns in .gitignore
Note that, although this solves the issue, it does not prevent from it popping
up again if new files get added to android-tools/ without explicitly excluding
them from .gitignore.
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Remove recipe of assimp which exists in oe-core.
* Patch 0001-Fix-build-on-big-endian-architectures.patch has been merged
in 4.0.0
* Dependencies 'boost' and 'virtual/libgl' are dropped which needded by
ASSIMP_BUILD_SAMPLES but ASSIMP_BUILD_SAMPLES is disabled by default
* Sync up other parts such as DESCRIPTION with the recipe in oe-core
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Since gobject-introspection convert to python3 in oe-core,
incorrectly assigning PYTHON_PN with python caused host python3
loaded.
On centos7, built with qemuarm failed:
...
arm-poky-linux-gnueabi-gcc: error: unrecognized argument in option
'-mtune=generic'
..
arm-poky-linux-gnueabi-gcc: error: unrecognized command line option
'-m64'
Traceback (most recent call last):
File "/usr/lib64/python3.4/distutils/unixccompiler.py", line 126, in
_compile
...
The native python3 should be loaded.
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Use github for SRC_URI
Reorder the linker flags so that we can remove disabling --as-needed
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
1) Upgrade openldap from 2.4.44 to 2.4.45
2) Delete openldap-CVE-2017-9287.patch, since it is integrated upstream.
3) License checksum changed, since the copyright years were updated.
Signed-off-by: Huang Qiyu <huangqy.fnst@cn.fujitsu.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
If "--with-pidfile-dir" configure option is not specified, then the syslog-ng
configure script assigns PID directory to /var/lib/syslog-ng; which is
where the PID file of syslog-ng process is created. But the syslog-ng
initscript (/etc/init.d/syslog-ng) uses /var/lib/syslog-ng/
as PID directory, this results in not stopping/restarting syslog-ng daemons
when it was requested to do so.
For example:
-- snip --
root@qemux86:~# /etc/init.d/syslog status
/usr/sbin/syslog-ng (pid 345 344) is running...
root@qemux86:~# /etc/init.d/syslog stop
Stopping syslog-ng:.
root@qemux86:~# /etc/init.d/syslog status
/usr/sbin/syslog-ng (pid 345 344) is running...
root@qemux86:~#
-- snip --
Hence set --with-pidfile-dir to "/var/run/syslog-ng", so that syslog-ng daemon
can be stopped/restarted via initscript.
Signed-off-by: Jagadeesh Krishnanjanappa <jkrishnanjanappa@mvista.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
- Fix compile failure against musl C library
- Fixed when compile with "-Wformat -Wformat-security -Werror=format-security"
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Tell Meson that even if it looks like the architectures are compatible, not to
bother executing binaries: differences in tunes, C libraries and so on mean
binaries may not work.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
