Since commit [libdevmapper/lvm2: force recipe libdevmapper to populate
sysroot only] applied, if recipe DEPENDS on libdevmapper, we have to
add lvm2 to DEPENDS
Remove duplicated bash from RDEPENDS
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Previously, in order to fix circular dependencies, we split
libdm from lvm2 as a new recipe libdevmapper, and as running result
proved, we need to keep both of them built with the same configuration
option
With new configuration options added, it is hard to keep them the same,
so we make recipe libdevmapper to populate libdevmapper sysroot only,
and make lvm2 to provide package libdevmapper.
In this situation, option differ does not affect header and library API
between lvm2 and libdevmapper, although lvm2 and libdevmapper are different
at build time, but there is only one lvm2 at run time
Side effect:
- If one recipe DEPENDS on libdevmapper, it has to add libdevmapper to
its RDEPENDS or add lvm2 to DEPENDS, otherwise, there will be a
[file-rdeps] or [build-deps] QA warning
- It has to add `PREFERRED_RPROVIDER_libdevmapper = "lvm2"' to local.conf
to workaound 'Multiple providers' NOTE
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
The package does use autotools based build systemd but it comes with a
bit of a twist. It explicitly disallows running autoheader on the source
tree. This can apparently be easilt accomplished with exclude parameter
passed via EXTRA_AUTORECONF.
This patch also adds PACKAGECONFIG option allowing the user to toggle
threading support in the library. Not sure how useful it is, but upstream
does offer such option, it works, and its not a lot of work to expose it
via PACKAGECONFIG.
Another improvement is related to package optimization handling.
mxml has a nasty habit of disregarding optimization flags passed by the
user. It always appends -Os -g to CFLAGS in its configure script. Its
a bit suprising to find out DEBUG_BUILD=1 in local.conf has no impact on
libraries produced by mxml recipe. This can be fixed with a simple sed
script.
Sed is also used to enable verbose compilation output. Having full
compiler invocations visible in the logs is generally useful, but
more importantly it allows some of the QA checks to their work.
While doing this cleanup I've also renamed the recipe from libxml_git.bb
to libxml_3.0.bb. The git keyword suggests this recipe is for some
random git snapshot, while in reality its for latest stable upstream
release of the package.
Last but not least this patch removes the bin package. Current versions of
mxml no longer provide any tools. The package was empty.
Signed-off-by: Piotr Tworek <tworaz@tworaz.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
While systemd enable cryptsetup (PACKAGECONFIG_append_pn-systemd = ' cryptsetup'),
there is a circular dependencies:
systemd -> systemd -> cryptsetup -> libdevmapper -> udev (PREFERRED_PROVIDER_udev = systemd)
The libdevmapper was used to break circular dependencies,
it does not require udev, refer commit [698d512 lvm2: Add a
PACKAGECONFIG for udev support] to move udev related from
lvm2.inc to lvm2.bb
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
ADE Framework is a graph construction, manipulation, and processing
framework. ADE Framework is suitable for organizing data flow
processing and execution.
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
clinfo is a simple command-line application that enumerates all possible
(known) properties of the OpenCL platform and devices available on the
system.
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
It was no longer building and will be removed in the next
upstream version.
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
cli11 is header-only and the main package is empty. Remove the
dependency to cli11 from cli11-dev so cli11-dev can be included in an
SDK via IMAGE_INSTALL, RDEPENDS, and the like.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Commit c6e963f9e ("lvm2: Add RDEPEND on lvm2 to lvm2-udevrules") added
a package dependency due to lvm2-udevrules needs dmsetup, however
dmsetup was moved to libdevmapper in commit 269d009a81 ("lvm2:
libdevicemapper package needs udev rules and dmsetup"), so this
dependency should be only for libdevmapper instead of the full package.
With the current implementation, a package that has a dependency with
lvm2-udev rules will include also many unnecessary packages like lvm2,
lvm2-scripts, etc. and their dependencies.
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
The unixODBC package no longer includes a driver for MySQL (that lives
elsewhere) so there is no reason to require MySQL libs at build time.
It does, however, want to build against readline, which was being pulled
in transitively by mysql5 before.
Signed-off-by: Dan Callaghan <dan.callaghan@opengear.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
This is a bug-fix release addressing the following issues:
- fix GCC8 warnings when building python bindings
- fix a bug in gpioinfo where lines used by the kernel without any
consumer string were reported as unused
- fix a bug in line::event_get_fd() in C++ bindings where the exception
wouldn't be thrown in error cases due to a missing 'throw' keyword
- fix a couple issues with the test-suite
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
This file is non-existent on non-x86 build hosts
Fixes
ppt.c:8:10: fatal error: sys/io.h: No such file or directory
#include <sys/io.h>
^~~~~~~~~~
Signed-off-by: Khem Raj <raj.khem@gmail.com>
It doesn't provide a way to enable or disable gobject-introspection
support for poppler. Toggle it respect to variable GI_DATA_ENABLED.
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Adjust the basename-include patch to deal with upstream change
of introducing goo/gbasename.cc.
Drop 0001-glib-CMakeLists.txt-Add-libpoppler-to-link-along-wit.patch
since that was done upstream in:
9481c473 glib: pass poppler to gir libs.
Signed-off-by: Randy MacLeod <Randy.MacLeod@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Unfortunately itstoll build system produces script file which contains
absolute path to python3 binary used during packaging. In case of OE
this refers to recipe-sysroot-native directory. This obviously make the
resulting script unusable for both target and native builds. This
problem was patched in OE using sed after the script file was installed.
The sed line replaced one absolute path with /usr/bin/python3. This
seems to work fine for target builds but is still not good enough for
native targets.s
The main problem with native builds is we can't mix host and OE provided
python3 when recipe using the tool inherits python3native bbclass. This
bbclass exports _PYTHON_SYSCONFIGDATA_NAME="_sysconfigdata" which is
likely to break host python3.
To make sure itstool is usable on both build host and target replace
/usr/bin/python3 absolute path with /usr/bin/env python3.
Signed-off-by: Piotr Tworek <tworaz@tworaz.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
The gnome class pulls in usually unwanted dependencies like gconf.
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
From: Anatol Belski <anbelski@microsoft.com>
The libctmalloc-minimal is what is useful for C++ apps if only the
improved allocator is needed. The gperftools package stays otherwise
fully backward compatible as it's going to depend on libtcmalloc-minimal.
The libtcmalloc-minimal can be installed separately, which allows for
more flexibility and for cleaner image contents.
The discussion on the patch
https://patchwork.openembedded.org/patch/163234/ has shown, that a
separate package for other libs might have less sense in this case, so
it's not done. The patch 163234 can be abandoned in favor of this for
now.
Signed-off-by: Anatol Belski <anbelski@microsoft.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
There have been a couple of releases since we last upgraded, including
a number of additional features. Changelog is here:
https://github.com/npat-efault/picocom/releases
Also tweak SUMMARY - I don't think this has been 20k in a while; with
this upgrade the binary alone went from 60k to 70k.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
If the local.conf has set 'USE_NLS = "no"' the build of libbytesize
will fail with the following message:
| configure: error: in `/build/tmp/work/core2-64-poky-linux/libbytesize/2.0-r0/build':
| configure: error: gettext not found
| See `config.log' for more details
| NOTE: The following config.log files may provide further information.
| NOTE: /build/tmp/work/core2-64-poky-linux/libbytesize/2.0-r0/build/config.log
| ERROR: configure failed
The issue is that libbytesize needs to full version of gettext in
order to configure properly, and not the minimal version which is
provided by default when building with 'USE_NLS = "no"'.
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
The workaround for gcc 9 broke building with gcc 8:
cc1plus: error: -Werror=deprecated-copy: no option -Wdeprecated-copy
cc1plus: error: -Werror=redundant-move: no option -Wredundant-move
ninja: build stopped: subcommand failed.
-Werror is often problematic in distribution builds,
disable it globally and permanently.
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
- Rebase use-sh-on-path.patch
- Drop 0001-fix-compile-failed-with-libc-musl.patch which
upstream has merged it
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
- Depends on libpcre2 since upstream commit [1c7e40a Port to pcre2]
- Remove PACKAGECONFIG python2 since upstream commit
[895baa3 Remove Python 2 support]
- Fix out of tree build failure
- RDEPENDS on python3-core rather than python3
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
The licence changed to GPLv2.
The fix from 0001-adjust-signature-of-main.patch is applied upstream.
joe 3.1 segfaults on aarch64, the update fixes this.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
after commit https://git.openembedded.org/openembedded-core/
commit/meta/classes/ptest.bbclass?id=b47194b57d94260b4e6438c5bf74914027f0b520
package ${PN}-ptest will depend on ${PN} by default,
but for uthash, ${PN} is empty package, remove it from dependency
to avoid image do rootfs failure since nothing provides error.
Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Update both opencv and opencv_contrib to 4.1.0 and refresh all patches.
Remove tiny-dnn from SRC_URI as it is not picked up by opencv cmake
build system. This also fixes a QA warning about using unstable github
archive.
Also update FILES_* to capture new paths which uses new directory name
"opencv4" instead of "OpenCV".
Signed-off-by: Chin Huat Ang <chin.huat.ang@intel.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Use libxml2 of oe-core to replace python3-lxml of meta-python
since meta-oe does not depend on meta-python
Set RDEPENDS_${PN}_class-native to fix `Missing or unbuildable
dependency chain was: ['meta-world-pkgdata', 'gnome-desktop3',
'itstool-native', 'libxml2-python-native']'
Apply changed from v3
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Use libxml2 of oe-core to replace python3-lxml of meta-python
since meta-oe does not depend on meta-python
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
