Christoph Ortner

Why do you need this? The compiler will tell you that it isn't implemented. But even if you insist you can just use ::Any

We need to depend on AtomsBase to share the functions that we use to access information stored in Atoms or Structures ie systems of particles. The interface is defined via...

- Maybe the documentation can be improved. - `Atoms` is not part of the expected interface as far as I understand, but an implementation detail? I could be wrong.

I've said this many times in many discussions and I'll say it again : in the absence of multiple inheritence (but to some extent even *with*) any sort of `AbstractSomethingOrOther`...

Maybe this now goes back to the #43 discussion. Personally, I would really prefer [something like this](https://github.com/ACEsuit/ACE.jl/blob/main/src/states.jl)

> We may still want to hold onto the data field, though, since otherwise ChemicalSpecies would essentially just be a rebrand of Element from PeriodicTable.jl I would want the species...

btw, the point about wrapping an integer is simply to distinguish the atomic number from an integer. There is no arithmetic on atomic numbers, one may want to dispatch on...

I don't think there has to be a compromise. If people are happy with the following idea, I can transfer this in from `JuLIP`: ```julia struct ChemicalSymbol # or ChemicalElement...

My point is that it does matter. If you store them as symbols then anything wrapping that will not be `isbits` and that can have real performance implications. The point...

Yes that’s essentially what I’m proposing. EDIT : above was a quick response - slightly longer : I suppose we can debate whether there is a point in having `ChemicalSymbol`...