Bump ndarray from 0.15.3 to 0.16

[Dirreke]

Blocked by jturner314/ndarray-npy#73

[Dirreke]

Should I rename it to into_shape_with_order, which is the same as ndarray(rust-ndarray/ndarray#1310)

tract/data/src/tensor.rs

Line 518 in fedab8d

pub fn into_shape(mut self, shape: &[usize]) -> TractResult<Tensor> {

[kali]

Should I rename it to into_shape_with_order, which is the same as ndarray(rust-ndarray/ndarray#1310)

tract/data/src/tensor.rs

Line 518 in fedab8d

pub fn into_shape(mut self, shape: &[usize]) -> TractResult<Tensor> {

No reason to. Tensors in tract are always in "natural order" (row major, but generalized, axes with bigger strides come first).

[Dirreke]

Hi~ Shall we rerun CI?

[kali]

Hi~ Shall we rerun CI?

Sure, let's see where we are. :) But we need to address the ndarray-npy situation before moving further (as in merge).

[barakugav]

@Dirreke @kali ndarray-npy 0.9.1 was released, using ndarray 0.16, can we push this PR forward? looking forward to it

[Dirreke]

@kali Let's rerun CI. Thanks

[kali]

thanks for your contibution!