January 2025 rebuild using rattler-build

[Tobias-Fischer]

No description provided.

[Tobias-Fischer]

@peci1: There is the following error on linux-aarch64, can you please take a look?

 │ │ FAILED: CMakeFiles/cras_log_utils.dir/src/log_utils/nodelet.cpp.o 
 │ │ $BUILD_PREFIX/bin/aarch64-conda-linux-gnu-c++ -DROSCONSOLE_BACKEND_LOG4CXX -DROS_BUILD_SHARED_LIBS=1 -DROS_PACKAGE_NAME=\"cras_cpp_common\" -Dcras_log_utils_EXPORTS -I$SRC_DIR/build/devel/include -I$
 │ │ SRC_DIR/ros-noetic-cras-cpp-common/src/work/include -I$PREFIX/share/xmlrpcpp/cmake/../../../include/xmlrpcpp -I$PREFIX/include/eigen3 -fvisibility-inlines-hidden -fmessage-length=0 -ftree-vectorize -
 │ │ fPIC -fstack-protector-strong -fno-plt -O3 -pipe -isystem $PREFIX/include -fdebug-prefix-map=$SRC_DIR=/usr/local/src/conda/ros-noetic-cras-cpp-common-2.4.7 -fdebug-prefix-map=$PREFIX=/usr/local/src/c
 │ │ onda-prefix -DBOOST_ERROR_CODE_HEADER_ONLY -D__STDC_FORMAT_MACROS=1 -O3 -DNDEBUG -std=gnu++17 -fPIC -fdiagnostics-color=always -DXMLRPCPP_HAS_PRINTTO=1 -MD -MT CMakeFiles/cras_log_utils.dir/src/log_u
 │ │ tils/nodelet.cpp.o -MF CMakeFiles/cras_log_utils.dir/src/log_utils/nodelet.cpp.o.d -o CMakeFiles/cras_log_utils.dir/src/log_utils/nodelet.cpp.o -c $SRC_DIR/ros-noetic-cras-cpp-common/src/work/src/log
 │ │ _utils/nodelet.cpp
 │ │ $SRC_DIR/ros-noetic-cras-cpp-common/src/work/src/log_utils/nodelet.cpp: In constructor 'cras::NodeletLogHelper::NodeletLogHelper(const GetNameFn&)':
 │ │ $SRC_DIR/ros-noetic-cras-cpp-common/src/work/src/log_utils/nodelet.cpp:23:45: error: no match for 'operator=' (operand types are 'cras::NodeletLogHelper::GetNameFn' {aka 'std::function<const std::__c
 │ │ xx11::basic_string<char>&()>'} and 'cras::NodeletLogHelper::NodeletLogHelper(const GetNameFn&)::<lambda()>')
 │ │    23 |     this->getNameFn = [](){ return NOT_SET; };
 │ │       |                                             ^
 │ │ In file included from $BUILD_PREFIX/lib/gcc/aarch64-conda-linux-gnu/14.2.0/include/c++/functional:59,
 │ │                  from $SRC_DIR/ros-noetic-cras-cpp-common/src/work/include/cras_cpp_common/log_utils/nodelet.h:12,
 │ │                  from $SRC_DIR/ros-noetic-cras-cpp-common/src/work/src/log_utils/nodelet.cpp:12:
 │ │ $BUILD_PREFIX/lib/gcc/aarch64-conda-linux-gnu/14.2.0/include/c++/bits/std_function.h:531:9: note: candidate: 'template<class _Functor> std::function<_Res(_ArgTypes ...)>::_Requires<std::function<_Res
 │ │ (_ArgTypes ...)>::_Callable<_Functor>, std::function<_Res(_ArgTypes ...)>&> std::function<_Res(_ArgTypes ...)>::operator=(_Functor&&) [with _Res = const std::__cxx11::basic_string<char>&; _ArgTypes =
 │ │  {}]'
 │ │   531 |         operator=(_Functor&& __f)
 │ │       |         ^~~~~~~~

[peci1]

Could you please point me to the whole build log? I can't find it in the checks attached to this PR.

Anyways, building noetic packages with gcc 14 is a really really bad idea. They were designed to work with gcc 9. Do you at least set default language level to c++14?

[Tobias-Fischer]

@peci1 - I fixed the issues, they were just minor.

[peci1]

Thanks. I will probably merge the fixes upstream.

Please also note that magnetometer_compass has been restructured and it now depends on a few new packages like magnetic_model or compass_conversions.

[Tobias-Fischer]

@traversaro - any chance that you could please take a look at the Windows failure of rviz (probably related to python-qt-bindings? Hopefully should not be too bad to fix, only very minor (and unrelated) things have changed in the ros-noetic-rviz.patch. However, I did bump from vs2017 to vs2019, perhaps that made the difference?

For reference, error is in https://github.com/RoboStack/ros-noetic/actions/runs/12604251974/job/35130798301

[Tobias-Fischer]

Windows fails with:

ninja: error: '%PREFIX%/Library/lib/lz4.lib', needed by 'devel/bin/pcl_ros_tf.dll', missing and no known rule to make it

Which is due to: https://github.com/conda-forge/conda-forge-pinning-feedstock/pull/6808/files

We need conda-forge/vtk-feedstock#360 and pin accordingly.

[Tobias-Fischer]

Note: we need conda-forge/qt-main-feedstock#315 and conda-forge/gazebo-feedstock#236 first.

[Tobias-Fischer]

We need conda-forge/gtk3-feedstock#74 and conda-forge/pcl-feedstock#77

[Tobias-Fischer]

We'll also need https://github.com/conda-forge/emacs-feedstock migrated to harfbuzz=10

[Tobias-Fischer]

Hi @traversaro - I think we're at the same state as in ros-jazzy with Windows being the only remaining issue. Are you ok to take a look, please? Many thanks :)

[traversaro]

Hi @traversaro - I think we're at the same state as in ros-jazzy with Windows being the only remaining issue. Are you ok to take a look, please? Many thanks :)

Ok!

[traversaro]

I completed a local rebuild. The only thing for which I have the doubt is the jackal-* additional recipes, pixi run build is now working, but I am not sure if the actual CI would work.

[traversaro]

Argh, somehow I broke the plotjuggler patches by moving one portion (that I had to change on Windows) in ros-noetic-plotjuggler.unix.patch.

[Tobias-Fischer]

Argh, somehow I broke the plotjuggler patches by moving one portion (that I had to change on Windows) in ros-noetic-plotjuggler.unix.patch.

Not a problem - it's an odd file that has different line endings and/or tabs vs spaces which gave me a few headaches previously. Have fixed it up, and merging now. Thanks a lot for the great collaborations on the rebuilds @traversaro, I highly appreciate it :)