Tiziano Müller

... the other thing which would prevent this exact issue (at least for the missing set number) would be to make our parser more robust (factor them out such that...

@fstein93 there's a Python-based parser for our format in here: https://github.com/dev-zero/aiida-gaussian-datatypes/blob/develop/aiida_gaussian_datatypes/basisset/utils.py

in parts this is due to Make unable to properly handle spaces in paths, see also [this stackoverflow answer](https://stackoverflow.com/a/9838604/1400465). For the line in question we can make it work for...

Not sure, it seems the Makefile was prepared to also discover other directories containing prettified files, not sure what this is about: ``` ALL_PREPRETTY_DIRS = $(shell find $(SRCDIR) -type d...

in this specific case there's something we can do, especially to make sure the user doesn't accidentally nuke his home dir

broken rules and builds is one thing, but blaming `make` for a misguided `rm -rf` is not really the way to go

That's one thing, yes since we don't support building in a spaced directory at all. The other thing is that in this particular case `find` might return a result with...

the point here is about being able to define paths explicitly according to builtin paths, for people who like to avoid random overrides of `@INCLUDE`d paths due to different working...

There is also the possibility of cross-compiling a Windows binary from Linux with mingw (without MPI): https://github.com/cp2k/cp2k/blob/master/arch/Linux-x86-64-mingw64-minimal.sopt Hence you should be able to build CP2K with just the mingw stack...

The problem is that you have to know about it, otherwise you'll waste time on investigating why suddenly a unittest fails which worked before.