potential name conflict

[Roger-luo]

Thanks for making this package, and very nice talk! Just want to notify that there is https://github.com/JuliaPhysics/Lattices.jl already which is for defining lattices.

The name of this package might be a bit misleading given the existence of that package. It'd be easier for users to install this package if the name can be changed to a more distinguishable one.

[donm]

Thanks, Roger!

Yeah, it's true about the name. We thought about this too, and deep down I sort of knew that I was effectively drafting you (and maybe others) into a conversation to about names when I created the repo.

I'll tell you the thoughts that Cody and I had when we initially talked about the name, but please tell me what you think, too. Our thoughts were basically:

• We're using the word "lattice" as a name in the same way that Flux.jl, Gadfly.jl, or JuMP.jl use their names less-than-literally, like a brand name.
• Lattice.jl won't provide general purpose data structures or functionality for lattices (and two packages called Lattice.jl and Lattices.jl that both provided that functionality really would be ridiculously confusing).
• If you know the Julia package naming conventions, then hearing a package with a singular name like Lattice or Gadfly doesn't really make you think that they're about lattices or gadflies, so it seemed like packages named with the singular and plural version of a word could be okay

On the other hand, just because frequent users of Julia might not be confused by a combination of names like FluxML/Flux.jl and JuliaPhysics/Fluxes.jl (if that existed), that doesn't mean somebody coming from PyTorch to experiment with Flux wouldn't be put off or confused. So I definitely see the other side to it, too.

[Roger-luo]

Hi, I think it's not about the name convention. It's simply just because there is only one letter difference in the name, which potentially make things very confusing. A user will need to install Lattice to find out it's actually not Lattices. For the same reason I don't think we will use Fluxes if there is a package for it. But PhysicsFlux etc. to make things more explicit.