introduced tests for pymc.distributions.logprob._get_scaling_

[chritter]

Fixes #5515. We have implemented tests for pymc.distributions.logprob.get_scaling which include edge cases and a test with a model graph random variable. Please let us know if the test coverage is sufficient or if further tests are required.

Co-authored-by: @Icyshaman.

Depending on what your PR does, here are a few things you might want to address in the description:

• what are the (breaking) changes that this PR makes?
• important background, or details about the implementation
• are the changes—especially new features—covered by tests and docstrings?
• linting/style checks have been run
• consider adding/updating relevant example notebooks
• right before it's ready to merge, mention the PR in the RELEASE-NOTES.md

#DataUmbrella Sprint
@reshamas

[codecov]

Codecov Report

Merging #5544 (af2e5f9) into main (53ad689) will decrease coverage by 2.55%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##             main    #5544      +/-   ##
==========================================
- Coverage   89.21%   86.66%   -2.56%     
==========================================
  Files          75       75              
  Lines       13846    13846              
==========================================
- Hits        12353    11999     -354     
- Misses       1493     1847     +354     

Impacted Files	Coverage Δ
pymc/smc/smc.py	18.47% <0.00%> (-77.92%)	⬇️
pymc/smc/sample_smc.py	16.77% <0.00%> (-66.45%)	⬇️
pymc/distributions/simulator.py	46.52% <0.00%> (-40.98%)	⬇️
pymc/backends/report.py	90.29% <0.00%> (-0.75%)	⬇️
pymc/step_methods/metropolis.py	82.55% <0.00%> (-0.43%)	⬇️
pymc/distributions/logprob.py	97.45% <0.00%> (+0.84%)	⬆️

[michaelosthege]

Looking quite good.
I'm not familiar with _get_scaling so I can't tell if the numbers are correct, but I assume that they are.

Can you add match="part of the message" to the pytest.raises(...)? That'll make sure that it's testing the right errors.

[michaelosthege]

@ricardoV94 @chritter can you check/fix/apply the suggestions I commented? Let's get this over the finish line

[michaelosthege]

Suggested change

	with pytest.raises(TypeError):
	with pytest.raises(TypeError, match="expected int or list of ints"):

[michaelosthege]

Suggested change

	with pytest.raises(ValueError):
	_get_scaling([1, 2, 5, Ellipsis, Ellipsis], (2, 3), 2)
	with pytest.raises(ValueError):
	_get_scaling([1, 2, 5, Ellipsis], (2, 3), 2)
	with pytest.raises(ValueError, match="Double Ellipsis"):
	_get_scaling([1, 2, 5, Ellipsis, Ellipsis], (2, 3), 2)
	with pytest.raises(ValueError, match="scalings is bigger that ndim"):
	_get_scaling([1, 2, 5, Ellipsis], (2, 3), 2)

[michaelosthege]

Suggested change

	with pytest.raises(ValueError):
	with pytest.raises(ValueError, match="scalings is bigger that ndim"):

[michaelosthege]

Suggested change

	with pytest.raises(TypeError):
	with pytest.raises(TypeError, match="expected int or list of ints"):

[michaelosthege]

I rebased it to fix git conflicts.

Will merge once the tests pass.