Open JoshuaSBrown opened 4 years ago
I'm not sure I agree with that, as all machine files do is record settings we pass to cmake
at configure time. Can you give an example of functionality you would like to abstract?
Take a look at the Darwin config file, I'm working on the config file for RZAnsel and have used a lot of the same logic. Standardizing the names of similar variables would also be a good idea.
Oh I see. Yes, because you're trying to be extremely general, that file is a rats nest. The machine config files I use personally I much smaller.
I would ask that the raw machine config files continue to work, even if some extra functionality is provided somewhere. It's very convenient for me to, e.g., throw together a file for my laptop or my personal setup on some HPC system.
Ya, I'm not talking about how the logic is handled in pathenon/CMakeLists.txt but simply packaging some of the more useful logic into functions. Perhaps it could be placed in parthenon/cmake/MachineConfig.cmake.
Oh I see. Yeah, that makes sense to me.
@AndrewGaspar I'm changing the title because I think this is more general than just LANL's HPC configs.
As more Machine Config files are added it makes sense to abstract out the common functionality.