Michael F. Herbst

indeed, but deamonmode is also tricky for other reasons (when should the daemon be shut down?) For the first version we will be stupid and recompile DFTK from scratch in...

Yes that's actually pretty much what we do right now anyway. Probably one should make it into a ToggableTimer package. Regarding the package, it definitely superseeds https://github.com/JuliaMolSim/DFTK.jl/blob/master/src/common/asserting.jl though

Theoretically yes, but since the original idea was to not pollute the stacktrace in case it is disabled then no (unless you want to compile each function twice, with and...

That's not fully true in the case of toggable assertions. [This line](https://github.com/MasonProtter/ToggleableAsserts.jl/blob/master/src/ToggleableAsserts.jl#L13) is always there even if assertions are turned off. The compiler will eventually remove it if assertions are...

I don't think we did in this context and I agree, it nicely abstracts a few of the common patterns we face. Sounds like a good candidate to try.

I agree. I looked at them a while back, but then what made me drop it was the bad spec. ONCVPSP are better specified ...

I disagree about the backref. I know we had it in the original design and later killed it for some reason, so I don't think reeintroducing it will be any...

Or of also storing the cartesian version of the G-vectors in the k-points

Model feels like the better solution too me even though ideally I would only want it to have the part that defines the lattice and the atoms, because a single...

Actually in this case it does. We use the combination of atoms and lattice a lot and I think it makes sense to add some nice support functions for this...