[Debian-med-packaging] Bug#1087010: prime-phylo:FTBFS:build failure (Imported target "MPI::MPI_Fortran" includes non-existent path )

[Aurelien Jarno]

control: reassign -1 openmpi-bin
control: forcemerge 1086209 -1
control: affects -1 prime-phylo

Hi,

On 2024-11-08 11:45, Yue Gui wrote:
> Source: prime-phylo
> Version:  1.0.11-13
> Severity: serious
> Tags: FTBFS, patch
> User: debian-riscv at lists.debian.org
> Usertags: riscv64
> X-Debbugs-Cc: debian-riscv at lists.debian.org
> 
> Dear prime-phylo Maintainer,
> The package prime-phylo build failed on all architectures except amd64. The
> crucial buildd log below:
> ```

Thanks for looking at build failures.

> -- Found MPI_C: /usr/lib/riscv64-linux-gnu/openmpi/lib/libmpi.so
> (found version "3.1")
> -- Found MPI_CXX: /usr/lib/riscv64-linux-gnu/openmpi/lib/libmpi.so
> (found version "3.1")
> CMake Error in /<<PKGBUILDDIR>>/obj-riscv64-linux-gnu/CMakeFiles/CMakeScratch/TryCompile-RHCg58/CMakeLists.txt:
>   Imported target "MPI::MPI_Fortran" includes non-existent path
> 
>     "/usr/lib/x86_64-linux-gnu/fortran/gfortran-mod-15/openmpi"
> 
>   in its INTERFACE_INCLUDE_DIRECTORIES.  Possible reasons include:
> 
>   * The path was deleted, renamed, or moved to another location.
> 
>   * An install or uninstall procedure did not complete successfully.
> 
>   * The installation package was faulty and references files it does not
>   provide.
> 
> 
> 
> CMake Error at /usr/share/cmake-3.30/Modules/FindMPI.cmake:1282 (try_compile):
>   Failed to generate test project build system.
> Call Stack (most recent call first):
>   /usr/share/cmake-3.30/Modules/FindMPI.cmake:1301 (_MPI_try_staged_settings)
>   /usr/share/cmake-3.30/Modules/FindMPI.cmake:1649 (_MPI_check_lang_works)
>   src/cxx/CMakeLists.txt:19 (find_package)
> 
> ```
> The full buildd log is here:
> https://buildd.debian.org/status/fetch.php?pkg=prime-phylo&arch=riscv64&ver=1.0.11-13&stamp=1730649664&raw=0
> My solution to this issue:
>   The configure failed because MPIFORTRAN was not set to the correct path.
> My solution is to properly set the path of MPIFORTRAN in the CMakeLists
> based on the current architecture. I have tested this method, and it is
> very effective. The debpatch is in the attachment. This patch can
> automatically set MPIFORTRAN to the correct path based on the current
> architecture. Please let me know whether this solution can be accepted.

I would argue that it would be better to get openmpi fixed to not use a
x86-64 hardcoded path in the package. This has actually been already
reported as bug #1086209. I am therefore reassigning and merging the
bugs there.

Regards
Aurelien

-- 
Aurelien Jarno                          GPG: 4096R/1DDD8C9B
aurelien at aurel32.net                     http://aurel32.net