ndslice: concatenate/vstack/hstack [$15 awarded]

[wilzbach]

Idea: Join multiple ndslices to a bigger ndslice
Maybe sth. like:

auto a = iotaSlice(2, 4);
auto b = 2.repeat.sliced(2, 4);
a.concatenate(b);

(Forum discussion)

---

The $15 bounty on this issue has been claimed at Bountysource.

[9il]

next week plan:

• selection.elementsMap - shell with opIndex, length and (optional) save methods.
• selection.concatenated - concatenation without allocation based on elementsMap
• slice.elements / slice.makeElements - allocates a common array composed of slice elements.
• slice.concatenation / ``slice.makeConcatenation` - multidimensional concatenation with allocation

[wilzbach]

I like the Api

1. probably a good opportunity to think more about a public isSlice method?
2. We shouldn't forgot there are multiple ways to stack (see below the way NumPy does it), but maybe we can have this as template argument?

http://docs.scipy.org/doc/numpy-1.10.1/reference/generated/numpy.vstack.html
http://docs.scipy.org/doc/numpy-1.10.1/reference/generated/numpy.concatenate.html#numpy.concatenate
http://docs.scipy.org/doc/numpy-1.10.1/reference/generated/numpy.stack.html#numpy.stack
http://docs.scipy.org/doc/numpy-1.10.1/reference/generated/numpy.hstack.html#numpy.hstack
http://docs.scipy.org/doc/numpy-1.10.1/reference/generated/numpy.dstack.html#numpy.dstack

[9il]

but maybe we can have this as template argument

yes

[Marenz]

Seeing that this was created May 15th and reading "next week plan" leads me to the suspicion that the issue might have been forgotten. So I figure a minor ping might be in order. ping

[9il]

@Marenz GLAS and ndslice.algorithm just have higher priority.

[Marenz]

Sure, no pressure, I just pinged in case it was forgotten ;)
Thanks for the work!

[9il]

@Marenz Thank you for the first Mir bounty! Currently we have issue with bountyscore, because Mir project is registered for DlangScience instead of Mir team. I hope we will fix it #345 soon