Pacman -U mingw-w64-x86_64-fpm-0.1. install this package run one of the following: conda install -c nvidia cuda conda install -c "nvidia/label/cuda-11.3.0" cuda conda install -c "nvidia/label/cuda-11.3.1" cuda conda install -c "nvidia/label/cuda-11.4.0" cuda conda install -c "nvidia/label/cuda-11.4.1" cuda conda install -c "nvidia/label/cuda-11.4.2" cuda conda install -c "nvidia/label/cuda-11.4.3" cuda conda install -c "nvidia/label/cuda-11.4.4" cuda conda install -c "nvidia/label/cuda-11.5.0" cuda conda install -c "nvidia/label/cuda-11.5.1" cuda conda install -c "nvidia/label/cuda-11.5.2" cuda conda install -c "nvidia/label/cuda-11.6.0" cuda conda install -c "nvidia/label/cuda-11.6.1" cuda conda install -c "nvidia/label/cuda-11.6.2" cuda conda install -c "nvidia/label/cuda-11.7.0" cuda conda install -c "nvidia/label/cuda-11.7.1" cuda conda install -c "nvidia/label/cuda-11.8.0" cuda conda install -c "nvidia/label/cuda-12.0.0" cuda conda install -c "nvidia/label/cuda-12.0.1" cuda conda install -c "nvidia/label/cuda-12.1. One can build and install fpm using the following commands ( PKGBUILD file is in the working directory): MINGW_INSTALLS=mingw64 makepkg-mingw
To initialize your shell, run conda initThe PKGBUILD would look somewhat like this: _realname=fpm If using conda activate from a batch script, change your invocation to CALL conda.bat activate. What is your experience with packaging Fortran projects? Have you found a good solution for creating distributions for Windows?įor Windows distributing a MinGW packages might be the way to go. Packaging for OSX is particular difficult because you cannot statically link binaries like on Linux, distribution outside of a package manager therefore always seemed unfeasible for me.ĭistribution for Windows has been an issue so far, conda-forge offers either classic flang 5 with LLVM or GFortran 5 with MinGW, which left me stuck with a similar issue I faced with homebrew: modern Fortran projects don’t compile with this setup. I also explored homebrew for OSX a while ago, which always has been a dead-end for modern Fortran because of their GFortran 4.9 version requirement to build the bottled packages. I’m currently relying mostly on conda-forge to distribute Fortran programs, which does a great job to reliably create packages for Linux (x86_64, ppc64le and aarch64) and OSX (x86_64) based on GFortran 9 (there is also aarch64 for OSX available now, but I haven’t tried this toolchain yet). Yet I get plenty of requests for Windows and OSX versions. Especially proprietary operating systems like Windows and OSX have always been hard to support, because I don’t have those available in my local development environment.
Reliable software distribution is one of the major issues I faced when starting my open source projects.
0 kommentar(er)
