Ivan Pribec

You are absolutely correct about the two-step installation process, it would be worth updating the README here to points towards the KarypisLab GitHub repositories: - https://github.com/KarypisLab/METIS - https://github.com/KarypisLab/GKlib Also the...

Hello trimtrim, you are correct, the arrays fxadj and fadjncy assume C-style numbering in this example. The numflag dummy variable should be set to 0. Surprisingly, I got no error...

I agree that the need to explictly call the subroutine c_f_pointer to gain access to the values of the array is not very user friendly. Also the fact that the...

Using the concept of C descriptors it is possible to copy the adjacency graph directly into Fortran allocatable arrays. First I created a new file `mesh_fortran.c` in the `metis-5.1.0/libmetis` folder...

This only matters if you are using the `read_graph` or `write_graph` routines which are not thoroughly tested yet. If not, I would recommend simply copying `metis_interface.f90` into your project. Concerning...

Hello @14NGiestas, is the aim of this project to provide something similar to [LAPACK95](http://www.netlib.org/lapack95/lug95/)? A few times I have asked myself if Fortran stdlib should also export some sort of...

Ok, so the missing generic interface to BLAS is one of the issues. Without doubt, in `stdlib` we will need to agree on some form of LAPACK, FFTW, and BLAS...

I know CMake reasonably well so I'm not looking for guidance. Essentially one would write custom script to compile the CUDA parts (or vice-versa), or split the program into two...

> What was the driving motivation behind such a system design? I find the manual instantiation of templates absolutely terrible for _users_ of the language, and showing a seeming disregard...

> > What was the driving motivation behind such a system design? I find the manual instantiation of templates absolutely terrible for _users_ of the language, and showing a seeming...