Support for 0d ndarrays

[antimatter15]

This doesn't seem to work with 0d ndarrays:

var a = ndarray(new Float32Array(128*128))
ops.addseq(a.pick(0), 2)

Throws "SyntaxError: Unexpected token |="

It looks like cwise-compiler (maybe this is a bug which should be filed there instead) is assembling some code which includes a var statement with no argument (i.e. "var \n a |= 0").

[rreusser]

Hmm… I can see the argument either way. The amount of data is defined as the product of the size of the dimensions. So a shape of [2, 2] has four numbers. [5] has five numbers. [] has no data. So it's a no-op.

So maybe it's that it should do nothing and exit early.

[mikolalysenko]

Yeah, I think this is a bug in cwise. Should open an issue there as well.

[Jabher]

I've made a fix: scijs/cwise-compiler#4
already accepted, waiting for npm release to make it work. This issue can be closed right after the release

[Jabher]

@rreusser correct, but not strictly correct. Size of array is shapes.reduce((a,b) => a * b, 1) which means that <2x2> tensor, which is matrix, has size 4; <2> tensor, which is vector, has size 2. But <> tensor (here goes mind blown) is scalar of size 1. And it is totally fine to use tensor algebra against it.