kumquat-buildroot/package/ola/ola.mk

181 lines
4.4 KiB
Makefile
Raw Normal View History

################################################################################
#
# ola
#
################################################################################
OLA_VERSION = 0.10.2
OLA_SITE = https://github.com/OpenLightingProject/ola/releases/download/$(OLA_VERSION)
OLA_LICENSE = LGPLv2.1+ (libola, libolacommon, Python bindings), GPLv2+ (libolaserver, olad, Python examples and tests)
OLA_LICENSE_FILES = LICENCE GPL LGPL
OLA_INSTALL_STAGING = YES
OLA_AUTORECONF = YES
# util-linux provides uuid lib
ola: bump to 0.9.3 This update requires a few changes. First, good news, the patches that you guys have submitted to OLA have been merged, so we can drop those: ola-0001-fix-build-warning.patch -> ea375582b0bfee93d66608ffc807078ffc48e961 ola-0002-move-python-sub-check-to-configure.ac.patch -> 673a7602a6bf7a6aa8a8461ebd9362d59f6e21df ola-0003-fix-check-for-python-module-for-cross-compilation.patch -> b51b48be81ec38bc7d1229be0c7d3189c5ddbafa Less good news: OLA now builds a small protoc wrapper (ola_protoc) that is built and executed at compile-time on the host. If we don't change anything, ola_protoc is built with the target toolchain and therefore can't run on the host. Explanation for ola_protoc is here To solve this, I created a package host-ola, which builds and provides an ola_protoc for the host. It tries to disable as much as possible of things that we won't need at configure time. Only ola_protoc is built and installed so it's not that long. The change has been built-tested only. [1] https://github.com/OpenLightingProject/ola/blob/master/protoc/ola-protoc.cpp#L20 [Thomas: - add HOST_OLA_DEPENDENCIES variable, so that host-ola doesn't get all the dependencies of the target ola package: certain target ola dependencies do not have corresponding host packages (and it doesn't make sense to have them as dependencies just to build ola_protoc) - improve the commit log to mention which upstream commits correspond to our patches.] Signed-off-by: Simon Marchi <simon.marchi@polymtl.ca> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
2014-12-08 01:39:39 +01:00
OLA_DEPENDENCIES = protobuf util-linux host-bison host-flex host-ola
OLA_CONF_OPTS = \
ac_cv_have_pymod_google_protobuf=yes \
--disable-gcov \
--disable-tcmalloc \
--disable-unittests \
--disable-root-check \
--disable-java-libs \
ola: bump to 0.9.3 This update requires a few changes. First, good news, the patches that you guys have submitted to OLA have been merged, so we can drop those: ola-0001-fix-build-warning.patch -> ea375582b0bfee93d66608ffc807078ffc48e961 ola-0002-move-python-sub-check-to-configure.ac.patch -> 673a7602a6bf7a6aa8a8461ebd9362d59f6e21df ola-0003-fix-check-for-python-module-for-cross-compilation.patch -> b51b48be81ec38bc7d1229be0c7d3189c5ddbafa Less good news: OLA now builds a small protoc wrapper (ola_protoc) that is built and executed at compile-time on the host. If we don't change anything, ola_protoc is built with the target toolchain and therefore can't run on the host. Explanation for ola_protoc is here To solve this, I created a package host-ola, which builds and provides an ola_protoc for the host. It tries to disable as much as possible of things that we won't need at configure time. Only ola_protoc is built and installed so it's not that long. The change has been built-tested only. [1] https://github.com/OpenLightingProject/ola/blob/master/protoc/ola-protoc.cpp#L20 [Thomas: - add HOST_OLA_DEPENDENCIES variable, so that host-ola doesn't get all the dependencies of the target ola package: certain target ola dependencies do not have corresponding host packages (and it doesn't make sense to have them as dependencies just to build ola_protoc) - improve the commit log to mention which upstream commits correspond to our patches.] Signed-off-by: Simon Marchi <simon.marchi@polymtl.ca> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
2014-12-08 01:39:39 +01:00
--disable-fatal-warnings \
--with-ola-protoc-plugin=$(HOST_DIR)/usr/bin/ola_protoc_plugin
ola: bump to 0.9.3 This update requires a few changes. First, good news, the patches that you guys have submitted to OLA have been merged, so we can drop those: ola-0001-fix-build-warning.patch -> ea375582b0bfee93d66608ffc807078ffc48e961 ola-0002-move-python-sub-check-to-configure.ac.patch -> 673a7602a6bf7a6aa8a8461ebd9362d59f6e21df ola-0003-fix-check-for-python-module-for-cross-compilation.patch -> b51b48be81ec38bc7d1229be0c7d3189c5ddbafa Less good news: OLA now builds a small protoc wrapper (ola_protoc) that is built and executed at compile-time on the host. If we don't change anything, ola_protoc is built with the target toolchain and therefore can't run on the host. Explanation for ola_protoc is here To solve this, I created a package host-ola, which builds and provides an ola_protoc for the host. It tries to disable as much as possible of things that we won't need at configure time. Only ola_protoc is built and installed so it's not that long. The change has been built-tested only. [1] https://github.com/OpenLightingProject/ola/blob/master/protoc/ola-protoc.cpp#L20 [Thomas: - add HOST_OLA_DEPENDENCIES variable, so that host-ola doesn't get all the dependencies of the target ola package: certain target ola dependencies do not have corresponding host packages (and it doesn't make sense to have them as dependencies just to build ola_protoc) - improve the commit log to mention which upstream commits correspond to our patches.] Signed-off-by: Simon Marchi <simon.marchi@polymtl.ca> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
2014-12-08 01:39:39 +01:00
HOST_OLA_DEPENDENCIES = host-util-linux host-protobuf
# When building the host part, disable as much as possible to speed up
# the configure step and avoid missing host dependencies.
HOST_OLA_CONF_OPTS = \
--disable-all-plugins \
--disable-slp \
--disable-osc \
--disable-uart \
--disable-libusb \
--disable-libftdi \
--disable-http \
--disable-examples \
--disable-unittests \
--disable-doxygen-html \
--disable-doxygen-doc \
--disable-fatal-warnings
ola: bump to 0.9.3 This update requires a few changes. First, good news, the patches that you guys have submitted to OLA have been merged, so we can drop those: ola-0001-fix-build-warning.patch -> ea375582b0bfee93d66608ffc807078ffc48e961 ola-0002-move-python-sub-check-to-configure.ac.patch -> 673a7602a6bf7a6aa8a8461ebd9362d59f6e21df ola-0003-fix-check-for-python-module-for-cross-compilation.patch -> b51b48be81ec38bc7d1229be0c7d3189c5ddbafa Less good news: OLA now builds a small protoc wrapper (ola_protoc) that is built and executed at compile-time on the host. If we don't change anything, ola_protoc is built with the target toolchain and therefore can't run on the host. Explanation for ola_protoc is here To solve this, I created a package host-ola, which builds and provides an ola_protoc for the host. It tries to disable as much as possible of things that we won't need at configure time. Only ola_protoc is built and installed so it's not that long. The change has been built-tested only. [1] https://github.com/OpenLightingProject/ola/blob/master/protoc/ola-protoc.cpp#L20 [Thomas: - add HOST_OLA_DEPENDENCIES variable, so that host-ola doesn't get all the dependencies of the target ola package: certain target ola dependencies do not have corresponding host packages (and it doesn't make sense to have them as dependencies just to build ola_protoc) - improve the commit log to mention which upstream commits correspond to our patches.] Signed-off-by: Simon Marchi <simon.marchi@polymtl.ca> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
2014-12-08 01:39:39 +01:00
# On the host side, we only need ola_protoc_plugin, so build and install this
# only.
HOST_OLA_MAKE_OPTS = protoc/ola_protoc_plugin
ola: bump to 0.9.3 This update requires a few changes. First, good news, the patches that you guys have submitted to OLA have been merged, so we can drop those: ola-0001-fix-build-warning.patch -> ea375582b0bfee93d66608ffc807078ffc48e961 ola-0002-move-python-sub-check-to-configure.ac.patch -> 673a7602a6bf7a6aa8a8461ebd9362d59f6e21df ola-0003-fix-check-for-python-module-for-cross-compilation.patch -> b51b48be81ec38bc7d1229be0c7d3189c5ddbafa Less good news: OLA now builds a small protoc wrapper (ola_protoc) that is built and executed at compile-time on the host. If we don't change anything, ola_protoc is built with the target toolchain and therefore can't run on the host. Explanation for ola_protoc is here To solve this, I created a package host-ola, which builds and provides an ola_protoc for the host. It tries to disable as much as possible of things that we won't need at configure time. Only ola_protoc is built and installed so it's not that long. The change has been built-tested only. [1] https://github.com/OpenLightingProject/ola/blob/master/protoc/ola-protoc.cpp#L20 [Thomas: - add HOST_OLA_DEPENDENCIES variable, so that host-ola doesn't get all the dependencies of the target ola package: certain target ola dependencies do not have corresponding host packages (and it doesn't make sense to have them as dependencies just to build ola_protoc) - improve the commit log to mention which upstream commits correspond to our patches.] Signed-off-by: Simon Marchi <simon.marchi@polymtl.ca> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
2014-12-08 01:39:39 +01:00
define HOST_OLA_INSTALL_CMDS
$(INSTALL) -D -m 0755 $(@D)/protoc/ola_protoc_plugin $(HOST_DIR)/usr/bin/ola_protoc_plugin
ola: bump to 0.9.3 This update requires a few changes. First, good news, the patches that you guys have submitted to OLA have been merged, so we can drop those: ola-0001-fix-build-warning.patch -> ea375582b0bfee93d66608ffc807078ffc48e961 ola-0002-move-python-sub-check-to-configure.ac.patch -> 673a7602a6bf7a6aa8a8461ebd9362d59f6e21df ola-0003-fix-check-for-python-module-for-cross-compilation.patch -> b51b48be81ec38bc7d1229be0c7d3189c5ddbafa Less good news: OLA now builds a small protoc wrapper (ola_protoc) that is built and executed at compile-time on the host. If we don't change anything, ola_protoc is built with the target toolchain and therefore can't run on the host. Explanation for ola_protoc is here To solve this, I created a package host-ola, which builds and provides an ola_protoc for the host. It tries to disable as much as possible of things that we won't need at configure time. Only ola_protoc is built and installed so it's not that long. The change has been built-tested only. [1] https://github.com/OpenLightingProject/ola/blob/master/protoc/ola-protoc.cpp#L20 [Thomas: - add HOST_OLA_DEPENDENCIES variable, so that host-ola doesn't get all the dependencies of the target ola package: certain target ola dependencies do not have corresponding host packages (and it doesn't make sense to have them as dependencies just to build ola_protoc) - improve the commit log to mention which upstream commits correspond to our patches.] Signed-off-by: Simon Marchi <simon.marchi@polymtl.ca> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
2014-12-08 01:39:39 +01:00
endef
# sets where to find python libs built for target and required by ola
OLA_CONF_ENV = PYTHONPATH=$(TARGET_DIR)/usr/lib/python$(PYTHON_VERSION_MAJOR)/site-packages
OLA_MAKE_ENV = PYTHONPATH=$(TARGET_DIR)/usr/lib/python$(PYTHON_VERSION_MAJOR)/site-packages
## OLA Bindings and Interface selections
ifeq ($(BR2_PACKAGE_OLA_WEB),y)
OLA_CONF_OPTS += --enable-http
OLA_DEPENDENCIES += libmicrohttpd
else
OLA_CONF_OPTS += --disable-http
endif
ifeq ($(BR2_PACKAGE_OLA_SLP),y)
OLA_CONF_OPTS += --enable-slp
else
OLA_CONF_OPTS += --disable-slp
endif
ifeq ($(BR2_PACKAGE_OLA_PYTHON_BINDINGS),y)
OLA_CONF_OPTS += --enable-python-libs
OLA_DEPENDENCIES += python python-protobuf
else
OLA_CONF_OPTS += --disable-python-libs
endif
## OLA Examples and Tests
ifeq ($(BR2_PACKAGE_OLA_EXAMPLES),y)
OLA_CONF_OPTS += --enable-examples
OLA_DEPENDENCIES += ncurses
else
OLA_CONF_OPTS += --disable-examples
endif
ifeq ($(BR2_PACKAGE_OLA_RDM_TESTS),y)
OLA_CONF_OPTS += --enable-rdm-tests
OLA_DEPENDENCIES += python-numpy
# needed as numpy builds some shared libraries and ola checks for
# numpy using a host python test program which fails with 'wrong ELF
# class'.
OLA_CONF_ENV = ac_cv_have_pymod_numpy=yes
else
OLA_CONF_OPTS += --disable-rdm-tests
endif
## OLA Plugin selections
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_E131),y)
OLA_CONF_OPTS += --enable-e131
else
OLA_CONF_OPTS += --disable-e131
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_ARTNET),y)
OLA_CONF_OPTS += --enable-artnet
else
OLA_CONF_OPTS += --disable-artnet
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_OPENDMX),y)
OLA_CONF_OPTS += --enable-opendmx
else
OLA_CONF_OPTS += --disable-opendmx
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_DUMMY),y)
OLA_CONF_OPTS += --enable-dummy
else
OLA_CONF_OPTS += --disable-dummy
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_ESPNET),y)
OLA_CONF_OPTS += --enable-espnet
else
OLA_CONF_OPTS += --disable-espnet
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_KINET),y)
OLA_CONF_OPTS += --enable-kinet
else
OLA_CONF_OPTS += --disable-kinet
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_OSC),y)
OLA_CONF_OPTS += --enable-osc
OLA_DEPENDENCIES += liblo
else
OLA_CONF_OPTS += --disable-osc
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_MILINT),y)
OLA_CONF_OPTS += --enable-milinst
else
OLA_CONF_OPTS += --disable-milinst
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_PATHPORT),y)
OLA_CONF_OPTS += --enable-pathport
else
OLA_CONF_OPTS += --disable-pathport
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_SANDNET),y)
OLA_CONF_OPTS += --enable-sandnet
else
OLA_CONF_OPTS += --disable-sandnet
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_SHOWNET),y)
OLA_CONF_OPTS += --enable-shownet
else
OLA_CONF_OPTS += --disable-shownet
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_STAGEPROFI),y)
OLA_CONF_OPTS += --enable-stageprofi --enable-libusb
else
OLA_CONF_OPTS += --disable-stageprofi
endif
ifeq ($(BR2_PACKAGE_OLA_PLUGIN_USBPRO),y)
OLA_CONF_OPTS += --enable-usbpro --enable-libusb
else
OLA_CONF_OPTS += --disable-usbpro
endif
$(eval $(autotools-package))
ola: bump to 0.9.3 This update requires a few changes. First, good news, the patches that you guys have submitted to OLA have been merged, so we can drop those: ola-0001-fix-build-warning.patch -> ea375582b0bfee93d66608ffc807078ffc48e961 ola-0002-move-python-sub-check-to-configure.ac.patch -> 673a7602a6bf7a6aa8a8461ebd9362d59f6e21df ola-0003-fix-check-for-python-module-for-cross-compilation.patch -> b51b48be81ec38bc7d1229be0c7d3189c5ddbafa Less good news: OLA now builds a small protoc wrapper (ola_protoc) that is built and executed at compile-time on the host. If we don't change anything, ola_protoc is built with the target toolchain and therefore can't run on the host. Explanation for ola_protoc is here To solve this, I created a package host-ola, which builds and provides an ola_protoc for the host. It tries to disable as much as possible of things that we won't need at configure time. Only ola_protoc is built and installed so it's not that long. The change has been built-tested only. [1] https://github.com/OpenLightingProject/ola/blob/master/protoc/ola-protoc.cpp#L20 [Thomas: - add HOST_OLA_DEPENDENCIES variable, so that host-ola doesn't get all the dependencies of the target ola package: certain target ola dependencies do not have corresponding host packages (and it doesn't make sense to have them as dependencies just to build ola_protoc) - improve the commit log to mention which upstream commits correspond to our patches.] Signed-off-by: Simon Marchi <simon.marchi@polymtl.ca> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
2014-12-08 01:39:39 +01:00
$(eval $(host-autotools-package))