building MeshLib on Debian 13

[fdarling]

Hello, I am trying to build MeshLib on Debian 13 (trixie) amd64. I did the following:

git clone --recurse-submodules https://github.com/MeshInspector/MeshLib.git
cd MeshLib
./scripts/build_thirdparty.sh

However it keeps complaining about missing dependencies. I started to install these through apt, but then I realized at least some of them are included as submodules in the thirdparty/ directory, such as the fmt library.

Why do I have to install packages such as libfmt-dev if thirdparty/fmt is included in the git repo? Is there a way to get it to "use" these built-in dependencies? Or are they only supposed to be used where packages are unavailable?

Also, using ./scripts/build_thirdparty.sh, is there a way to disable certain modules? Something similar to doing:

mkdir build
cd build
cmake .. -DMESHLIB_PYTHON_SUPPORT=OFF -DMESHLIB_BUILD_MRVIEWER=OFF -DMESHLIB_BUILD_MESHVIEWER=OFF -DMESHLIB_BUILD_MESHCONV=OFF -DMESHLIB_BUILD_SYMBOLMESH=OFF -DMESHLIB_BUILD_VOXELS=OFF -DMESHLIB_BUILD_EXTRA_IO_FORMATS=OFF -DMESHLIB_BUILD_MRCUDA=OFF

I'd appreciate some guidance!

[Grantim]

Hello!

In general we want to use as many native dependencies as possible, but on platforms that have incompatible versions or don't have required packages at all we use submodules.

Other thing is that we do not check previous statement during build but have predefined knowledge about platforms that we are testing on, and unfortunately, we don't test on debian (not sure if it will help but you can try running https://github.com/MeshInspector/MeshLib/blob/master/scripts/install_apt_requirements.sh).

Anyway, we have some work done to support vcpkg on linux, it will probably be useful on platforms that we don't support natively:

MeshLib/docker/rockylinux9-vcpkgDockerfile

Lines 21 to 43 in 2bfc5c7

	RUN git clone https://github.com/microsoft/vcpkg.git --branch $VCPKG_VERSION --single-branch && \
	cd vcpkg && \
	./bootstrap-vcpkg.sh -disableMetrics
	WORKDIR /root/vcpkg
	# replace the original source URL with a faster mirror's one
	RUN sed -i 's|https://ftp.gnu.org/|https://www.mirrorservice.org/sites/ftp.gnu.org/|' ports/vcpkg-make/portfile.cmake && \
	sed -i 's|https://ftp.gnu.org/|https://www.mirrorservice.org/sites/ftp.gnu.org/|' ports/gettext/portfile.cmake
	# build vcpkg packages
	COPY requirements/vcpkg-linux.txt MeshLib/requirements.txt
	COPY thirdparty/vcpkg/ports MeshLib/ports
	COPY thirdparty/vcpkg/triplets MeshLib/triplets
	ARG VCPKG_TRIPLET
	ENV VCPKG_TRIPLET=$VCPKG_TRIPLET
	RUN source /opt/rh/autoconf271/enable && \
	./vcpkg install --host-triplet=${VCPKG_TRIPLET} --overlay-triplets=MeshLib/triplets --overlay-ports=MeshLib/ports $(cat MeshLib/requirements.txt | tr '\n' ' ')
	RUN ./vcpkg export --host-triplet=${VCPKG_TRIPLET} --overlay-triplets=MeshLib/triplets --x-all-installed --raw --output=vcpkg --output-dir=/opt
	FROM nvidia/cuda:12.6.0-devel-rockylinux9 AS production

MeshLib/.github/workflows/build-test-linux-vcpkg.yml

Lines 141 to 156 in 2bfc5c7

	- name: Build
	run: ./scripts/build_source.sh
	env:
	MESHLIB_BUILD_RELEASE: ${{ fromJSON('["OFF", "ON"]')[matrix.config == 'Release'] }}
	MESHLIB_BUILD_DEBUG: ${{ fromJSON('["OFF", "ON"]')[matrix.config == 'Debug'] }}
	CMAKE_CXX_COMPILER: ${{ matrix.cxx-compiler }}
	MR_VERSION: ${{ inputs.app_version }}
	# options to be passed to cmake
	MR_CMAKE_OPTIONS: >
	-DMR_PLATFORM=Linux_vcpkg
	-DMR_CXX_STANDARD=${{ matrix.cxx-standard }}
	-DMR_PCH_USE_EXTRA_HEADERS=ON
	-DCMAKE_CUDA_HOST_COMPILER=/opt/rh/gcc-toolset-11/root/usr/bin/g++
	-DMESHLIB_BUILD_GENERATED_C_BINDINGS=${{ fromJSON('["OFF", "ON"]')[inputs.mrbind_c] }}
	-DMRVIEWER_NO_GTK=ON
	-DMRVIEWER_WITH_BUNDLED_CURL=ON

Also, using ./scripts/build_thirdparty.sh, is there a way to disable certain modules? Something similar to doing:

You can define additional MR_CMAKE_OPTIONS env to provide these flags to build script, or you can use cmake directly

MeshLib/scripts/build_source.sh

Lines 146 to 149 in 2bfc5c7

	cd build/Release
	cmake -S ../.. -B . -D CMAKE_BUILD_TYPE=Release ${MR_CMAKE_OPTIONS} $@ | tee ${logfile}
	cmake --build . -j ${NPROC} | tee ${logfile}
	cd ../..