How about just making calc.abs variadic, and equal to the euclidean norm? It makes sense for any number of arguments, as long as they're either all lengths or all numbers.

I like this idea, but I can see two problems: readability (it's not obvious to the uninformed reader what calc.abs(x, y, z) does), and discoverability. It may be better to create a separate calc.norm function and keep calc.abs separate, just for calculating absolute values. This would solve the readability problem, and also improve discoverability.

Additionally, we might as well go all the way and accept a positional argument to specify what p-norm to compute (not sure what the name would be though).