Incorrect data assignment with some multiple list indices

[davidhassell]

Incorrect data assignment with some multiple list indices

[davidhassell]

Note - Whilst cfdm never said or trapped otherwise, this only ever really worked when all list indices (when there are at least two) were strictly monotonically increasing. This is because this code was back-ported from cf-python, which prior to v3.14 did insist on this (it was a LAMA thing). There is no need for this restriction in cf-python>=3.14, nor in cfdm.

[sadielbartholomew]

@davidhassell: whilst I am reviewing the associated PR, I just wanted to ask for clarity of describing the bug here: the first and second examples seems a bit confusing and contradictory to me, I am wondering if you wrote down quite what you meant? Namely, did you mean to put that the second example is 'OK' rather than 'NOT OK' since it seems to set the elements to -1 as asked? And then the first example says is covers all lists strictly monotonically increasing but the indices aren't in that case?

I was thinking the following were more like what I am understanding you want to refer to:

Example 1 (all lists strictly monotonically increasing)

In [4]: d[[0, 3, 4], [1, 6, 7]].array
Out[4]: 
array([[0.034, 0.024, 0.029],
       [0.059, 0.009, 0.017],
       [0.036, 0.034, 0.013]])

In [5]: d[[0, 3, 4], [1, 6, 7]] = -1

In [6]: d[[0, 3, 4], [1, 6, 7]].array  # OK!
Out[6]: 
array([[-1., -1., -1.],
       [-1., -1., -1.],
       [-1., -1., -1.]])

Example 2 (one list not strictly monotonically increasing, but in this case it works)

In [7]: d = cfdm.example_field(0).data

In [8]: d[[0, 3, 4], [1, 7, 6]].array
Out[8]: 
array([[0.034, 0.029, 0.024],
       [0.059, 0.017, 0.009],
       [0.036, 0.013, 0.034]])

In [9]: d[[0, 3, 4], [1, 7, 6]] = -1

In [10]: d[[0, 3, 4], [1, 7, 6]].array  # OK!
Out[10]: 
array([[-1., -1., -1.],
       [-1., -1., -1.],
       [-1., -1., -1.]])

[davidhassell]

Good spot! Your right, and I'll edit the original post.

[sadielbartholomew]

Good spot! Your right, and I'll edit the original post.

Thanks for checking. It helps to know I am understanding correctly the issue I am meant to be reviewing on 🙂