fix sheBangs for python3

A lot a smalls things on msys2/mingw64 config :

• Fix shebang to python3
• port to mingw64 (msys2) : minor changes
• remove the old (unusable anymore luc.cmake)

Hi Luc,

Thanks for correcting the shebangs, I must have missed them!

There is already a msys2.cmake. Maybe you could use this one instead of creating msys2/mingw64?
Only complain I have with what you wrote is SET(WAVES_USE_MUMPS ON CACHE BOOL "" FORCE). It default to OFF as MUMPS is not usually present in MSYS environment.

What do you think?

for mumps, i just used the msys/mingw precompiled package (i did not compiled Romain's git one) and even if it is a mumps 5.3.4-1 it worked with waves & pfem ... => for me no reason not to use it ...

I had more problems with Eigen3 (system) and gmm (installed : no package) : i had to modify .bashrc to get them found by cmake (i still have to speak with Romain how to manage the "system" config for msys)

# lpx Configuration
export INCLUDE=/mingw64/include/eigen3/:/opt/gmm/include/
export PATH=~/bin:/opt/gmsh/bin:/opt/gmsh/lib:$PATH
export PYTHONPATH=C:/ProgramFiles/msys64/opt/gmsh/lib

I also had trouble in cmake while it looked for an icc (not installed) not gcc (even CXX and CC env variable did not worked), reason why they are defined in the cmake file (not understood)... finally, i did not solved the mpi problem (not took the time to) anyway, no proble, i can remove the config file

for mumps, i just used the msys/mingw precompiled package

Ok, I did not known they added a package...

I had more problems with Eigen3 (system) and gmm (installed : no package) : i had to modify .bashrc to get them found by cmake (i still have to speak with Romain how to manage the "system" config for msys)

That is also what I did (expect that I do it on the fly, when I have to recompile). I guess gmm is for Metafor since waves does not use gmm anymore.

I also had trouble in cmake while it looked for an icc (not installed) not gcc (even CXX and CC env variable did not worked), reason why they are defined in the cmake file (not understood)...

Never has this issue

finally, i did not solved the mpi problem (not took the time to) anyway

No worries, we know it works when you use MSMPI and that you compile mpi4py from sources

i can remove the config file

Lets merge the best of both. Thanks!

added 1 commit

• eca34777 - remove msys2/mingw64 config file

Compare with previous version

removed msys2-mingw64.cmake for gmm it is for pfem (not Metafor)

after a last check the mumps version is not the system one, but the 4.10 one available on the gitlab (i did not remeber i compiled it)

after a last check the mumps version is not the system one, but the 4.10 one available on the gitlab (i did not remeber i compiled it)

Still, the msys2 package exists (https://packages.msys2.org/package/mingw-w64-x86_64-mumps?repo=mingw64). So you can leave it as it is, I don't mind (I'll just have to update the doc I guess).

approved this merge request

I don't see any old_div removed

changed title from fix sheBangs for python3 and add mingw64 cmake configuration file to fix sheBangs for python3

changed the description

there were no old_div remaining in waves, but in pfem (i mixed the two as i worked on the two repos at the same time) => removed from description for mumps, libs are not named the same way that what is done in the mumps-4.10 repo => the findMumps would have to be fixed (and in that case, it would be great to make the work for all the platforms). It may also have porting operations (but i don't have time to do it)... So up to now, i would let it like it is ...

ok! I also need to test again msys/mingw on my PC

approved this merge request

merged