The CONTRIBUTING.md requests an (or multiple) SPDX identifier for GPL
licenses. But a lot of packages did use a different, non-SPDX style with a
"+" at the end instead of "-or-later".
Signed-off-by: Sven Eckelmann <sven@narfation.org>
fix Makefile chmod (644)
replace MD5SUM with HASH
add PKG_MIRROR_HASH when PKG_SOURCE_PROTO:=git
(PKG_SOURCE_PROTO:=svn tarballs are not reproducible for now)
Signed-off-by: Etienne Champetier <champetier.etienne@gmail.com>
-fhonour-copts is an OpenWrt/LEDE toolchain specific patch, forcing it
will make external toolchain fail to build this package. What we need
instead is a way to override CFLAGS from the different Makefile files so
rename the patch to illustrate that.
Fixes: 317c2469d9 ("libcanfestival: fix several small build process issues")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
This was suggested by @hnyman because hg is not a prerequisite in
LEDE or Openwrt, so using that should be avoided, if possible.
The tricky part is that the filename is appended to the URL,
thus we construct/use an URL where the appended filename does not hurt.
Signed-off-by: Michael Heimpold <mhei@heimpold.de>
This patch tries to address the following package issues:
- the upstream hg repository seems to contain some invalid timestamps,
at least the following warning are generated during package compilation
(only one example):
emcy.c: Timestamp out of range; substituting 2514-05-30 01:53:03.999999999
This can be fixed by touching all files after unpacking, thus tar is
now given the -m switch.
- one or more LEDE buildbot(s) do not have hg tool installed and thus are not
able to checkout the sources from upstream
This patch populates PKG_MIRROR_HASH in the hope that the buildbot's first
try is to download an already packed source tarball prior to checkout it out
from upstream.
- While at packaging from upstream, use the common infrastructure, i.e. use
a source date etc. This also results in tar.xz files and thus reduces the
disk footprint a little bit.
- during compilation the follwing warning could be observed:
cc1: note: someone does not honour copts correctly, passed 0 times
To silent down these warning, patch the Makefiles and pass the expected
option.
Important note: changes are only compile tested due to missing hardware at
hand.
Signed-off-by: Michael Heimpold <mhei@heimpold.de>
This should fix the build on LEDE buildbots which bail out with:
-snip-
...
! -d ./src/ ] || cp -fpR ./src/* /data/bowl-builder/arm_arm926ej-s/build/ (wrapped)
sdk/build_dir/target-arm_arm926ej-s_musl-1.1.15_eabi/libcanfestival-8bfe0ac0
Applying ./patches/001-sigval-ref-fix.patch using plaintext:
can't find file to patch at input line 3
Perhaps you used the wrong -p or --strip option?
The text leading up to this was:
--------------------------
|--- a/drivers/timers_unix/timers_unix.c
|+++ b/drivers/timers_unix/timers_unix.c
--------------------------
No file to patch. Skipping patch.
1 out of 1 hunk ignored
Patch failed! Please fix ./patches/001-sigval-ref-fix.patch!
...
-snap-
Reason is, that the tar ball created by hg checkout does not contain the
version string appended to the root source directory as expected by
default PKG_BUILD_DIR, so this patch adjusts the expected directory name.
Signed-off-by: Michael Heimpold <mhei@heimpold.de>
At the moment, LEDE buildbots are complaining with:
-snip-
...
libcanfestival/examples/TestMasterSlave/TestMasterSlave.c:50: undefined reference to `MasterMap1'
TestMasterSlave.o: In function `InitNodes':
...
-snap-
Since we are only interessted in the library itself, skip compilation
of the example code. This should both fix the build and speedup it
a little bit.
Signed-off-by: Michael Heimpold <mhei@heimpold.de>
The previous attempt to fix the build error was not successfull
and introduced needlessly a new patch. However, the configure
already provide the required functionality, use it instead.
This should now really solve the build.
Signed-off-by: Michael Heimpold <mhei@heimpold.de>
Sven Eckelmann
Etienne Champetier
Florian Fainelli
Michael Heimpold
Anton Glukhov