Fix invalid-sequence-index for call to scipy.fft.rfft

[clavedeluna]

Type of Changes

	Type
✓	🐛 Bug fix

Description

Fix invalid-sequence-index for call to scipy.fft.rfft.

Closes #8018

[clavedeluna]

I'd like feedback from maintainers on where I should add scipy as a test dependency

[Pierre-Sassoulas]

I think it's better to pinpoint the exact code construct that create the problem in scipy and then use that directly in the test. We won't have to install scipy, and the test will still be meaningful if the scipy code is refactored and the problematic code construct disappear.

[github-actions]

🤖 Effect of this PR on checked open source code: 🤖

Effect on django:
The following messages are now emitted:

1. subprocess-run-check:
   'subprocess.run' used without explicitly defining the value for 'check'.
   https://github.com/django/django/blob/ef85b6bf0bc5a8b194f0724cf5bbedbcee402b96/django/core/management/utils.py#L22
2. subprocess-run-check:
   'subprocess.run' used without explicitly defining the value for 'check'.
   https://github.com/django/django/blob/ef85b6bf0bc5a8b194f0724cf5bbedbcee402b96/django/core/management/utils.py#L172
3. subprocess-run-check:
   'subprocess.run' used without explicitly defining the value for 'check'.
   https://github.com/django/django/blob/ef85b6bf0bc5a8b194f0724cf5bbedbcee402b96/django/utils/autoreload.py#L274
4. subprocess-run-check:
   'subprocess.run' used without explicitly defining the value for 'check'.
   https://github.com/django/django/blob/ef85b6bf0bc5a8b194f0724cf5bbedbcee402b96/django/utils/version.py#L87

The following messages are no longer emitted:

1. subprocess-run-check:
   Using subprocess.run without explicitly set check is not recommended.
   https://github.com/django/django/blob/ef85b6bf0bc5a8b194f0724cf5bbedbcee402b96/django/core/management/utils.py#L22
2. subprocess-run-check:
   Using subprocess.run without explicitly set check is not recommended.
   https://github.com/django/django/blob/ef85b6bf0bc5a8b194f0724cf5bbedbcee402b96/django/core/management/utils.py#L172
3. subprocess-run-check:
   Using subprocess.run without explicitly set check is not recommended.
   https://github.com/django/django/blob/ef85b6bf0bc5a8b194f0724cf5bbedbcee402b96/django/utils/autoreload.py#L274
4. subprocess-run-check:
   Using subprocess.run without explicitly set check is not recommended.
   https://github.com/django/django/blob/ef85b6bf0bc5a8b194f0724cf5bbedbcee402b96/django/utils/version.py#L87

This comment was generated for commit b11d2d6

[clavedeluna]

I think it's better to pinpoint the exact code construct that create the problem in scipy and then use that directly in the test. We won't have to install scipy, and the test will still be meaningful if the scipy code is refactored and the problematic code construct disappear.

That makes sense, but I think I'll still need to at least test with numpy since the behavior happens when the objects are numpy arrays. I see some functional tests that do use numpy, but I don't see where numpy is installed?? @Pierre-Sassoulas

[Pierre-Sassoulas]

There are import from numpy but it's tested without installing numpy (it's simply an example using numpy). We would need a numpy astroid plugin if something is specific to numpy. We might need to discuss how we test c extension generically by providing the C code ourselves to be able to follow the same logic than m'y previous comment for c extensions.

[clavedeluna]

I've pushed the latest code I have but will block this until maintainers let me know about how to move forward with testing this.