Add test for mutating input arrays #8990 #24511

saranti · 2022-11-20T04:34:28Z

PR Summary

PR Checklist

Documentation and Tests

Has pytest style unit tests (and pytest passes)
Documentation is sphinx and numpydoc compliant (the docs should build without error).
New plotting related features are documented with examples.

Release Notes

New features are marked with a .. versionadded:: directive in the docstring and documented in doc/users/next_whats_new/
API changes are marked with a .. versionchanged:: directive in the docstring and documented in doc/api/next_api_changes/
Release notes conform with instructions in next_whats_new/README.rst or next_api_changes/README.rst

github-actions

Thank you for opening your first PR into Matplotlib!

If you have not heard from us in a while, please feel free to ping @matplotlib/developers or anyone who has commented on the PR. Most of our reviewers are volunteers and sometimes things fall through the cracks.

You can also join us on gitter for real-time discussion.

For details on testing, writing docs, and our review process, please see the developer guide

We strive to be a welcoming and open project. Please follow our Code of Conduct.

story645 · 2022-11-20T05:45:49Z

lib/matplotlib/tests/test_mutate_inputs.py

+pl.get_data()
+
+
+def test_1():


can you please give this test a more descriptive name? and if that test data is only used for this test, please put it in the function. If it's used for both, please put it in a fixture?

story645 · 2022-11-20T05:46:36Z

lib/matplotlib/tests/test_mutate_inputs.py

+    assert str(pl.get_data_3d()) == result1
+    assert str(pl.get_data()) == result2


why are these strings?

That was the only way i was able to get a matching comparison on the tuple. Is there a better way to do this?

You can do something like

np.all(p1.get_data() == result2)

where result2 is then an np-array.

You may have to do np.allclose instead of np.all as there may be floating-point effects, at least on result2

Thanks Oscar. Using np.allclose appears to return an assertion error on Ubuntu and Mac:

FAILED lib/mpl_toolkits/mplot3d/tests/test_axes3d.py::test_mutating_input_arrays_y_and_z - assert False is True + where False = <function allclose at 0x7f4c5b088ee0>((array([-0.07780068, -0.02919708, 0.02192821]), array([-0.04741606, -0.06321353, -0.07983062])), ((-0.07348146, -0.02919708, 0.01717111), (-0.04881992, -0.06321353, -0.07828444))) + where <function allclose at 0x7f4c5b088ee0> = np.allclose + and (array([-0.07780068, -0.02919708, 0.02192821]), array([-0.04741606, -0.06321353, -0.07983062])) = <bound method Line2D.get_data of <mpl_toolkits.mplot3d.art3d.Line3D object at 0x7f4c38963730>>() + where <bound method Line2D.get_data of <mpl_toolkits.mplot3d.art3d.Line3D object at 0x7f4c38963730>> = <mpl_toolkits.mplot3d.art3d.Line3D object at 0x7f4c38963730>.get_data = 1 failed, 8896 passed, 207 skipped, 13 xfailed, 7 xpassed, 5 warnings in 706.06s (0:11:46) = Error: Process completed with exit code 1.

I'm not really sure why this is happening.

The values are slightly different:

(array([-0.07780068, -0.02919708, 0.02192821]), array([-0.04741606, -0.06321353, -0.07983062]))

vs

((-0.07348146, -0.02919708, 0.01717111), (-0.04881992, -0.06321353, -0.07828444))

I am wondering if setting limits etc will somehow make it different?

oscargus · 2022-11-20T16:00:55Z

Please add the content of the tests as functions to https://github.com/matplotlib/matplotlib/blob/main/lib/mpl_toolkits/mplot3d/tests/test_axes3d.py instead.

oscargus

It may be easier to just do an image_comparison test. See

matplotlib/lib/mpl_toolkits/mplot3d/tests/test_axes3d.py

Lines 305 to 306 in c5ed8eb

    
           @check_figures_equal(extensions=["png"]) 
        
           def test_plot_scalar(fig_test, fig_ref):

for an example.

So do exactly the same operation on both fig_test and fig_ref (using different arrays for input data) and then mutate one of the input arrays.

This holds for both tests and will avoid the problem of figuring out the exact coordinates of the plotting results.

It can be good to make a comment in each test pointing to GH#8990 so that one understands where the test case comes from.

oscargus · 2022-11-21T09:07:33Z

lib/mpl_toolkits/mplot3d/tests/test_axes3d.py

+    pl.get_data_3d()
+    pl.get_data()


Suggested change

pl.get_data_3d()

pl.get_data()

Not required

Thanks for the help, the image comparison test worked fine on both of them. Now I just have that pr_clean test failing because the file I originally wrote the test on was added and deleted.

lib/mpl_toolkits/mplot3d/tests/test_axes3d.py

oscargus · 2022-11-21T09:14:30Z

lib/mpl_toolkits/mplot3d/tests/test_axes3d.py

+    pl.get_data_3d()
+    pl.get_data()
+
+    array = tuple


Better to remove the array call below.

oscargus · 2022-11-21T09:16:15Z

lib/mpl_toolkits/mplot3d/tests/test_axes3d.py

+    pl.get_data()
+
+    array = tuple
+    result1 = (array([1, 2, 3]), array([0., 0., 0.]), array([0., 0., 0.]))


Suggested change

result1 = (array([1, 2, 3]), array([0., 0., 0.]), array([0., 0., 0.]))

result1 = ((1, 2, 3),(0., 0., 0.), (0., 0., 0.))

allclose will check that the numerical values are close, so if using lists or tuples (or np.array) doesn't matter as long as the structure is correct. (Modulo that it may not be correct in my simplified version...)

Same on the next row.

oscargus · 2022-11-21T09:17:25Z

lib/mpl_toolkits/mplot3d/tests/test_axes3d.py

+    assert np.allclose(pl.get_data_3d(), result1) is True
+    assert np.allclose(pl.get_data(), result2) is True


Suggested change

assert np.allclose(pl.get_data_3d(), result1) is True

assert np.allclose(pl.get_data(), result2) is True

assert np.all(pl.get_data_3d() == result1)

assert np.allclose(pl.get_data(), result2)

For result1 the values are exact enough to compare directly. is True not required as assert is checking for True anyway.

oscargus · 2022-11-21T09:18:07Z

lib/mpl_toolkits/mplot3d/tests/test_axes3d.py

+    ax2.plot(x, V2, zs=2, zdir='y')
+    fig.draw_without_rendering()
+
+    result_x = "[0.  0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.  1.1 1.2 1.3" \


Please use numbers in tuples/lists here as well.

tacaswell · 2022-11-22T23:34:08Z

lib/mpl_toolkits/mplot3d/tests/test_axes3d.py

+
+# test cases came from GH#8990
+@check_figures_equal(extensions=["png"])
+def test_mutating_input_arrays_sin_3d(fig_test, fig_ref):


What is this testing that the other test is not testing?

It's testing that the second curve doesn't affect the first curve as reported here.

Edit: I just attempted a rebase but the added and deleted file hasn't been affected by it. It still won't pass the pr_clean test.

In that example the important change is mutating the input not that we plotted it a second time (if you go back to 3.5.2 and remove the second plot I expect it to be broken).

Ok. In that case, should we remove the second test entirely?

tacaswell · 2022-11-22T23:40:12Z

Thank you for your work on this @saranti ! Are you comfortable doing a rebase? This needs to be squashed down to one commit (we have a check that makes sure we do not add and then remove a file in a PR. This helps keep the repo size down, makes sure we do not accidentally commit a whole venv again, and provides a mild security check that no one sneaks a malicious file in!).

tacaswell · 2022-11-23T17:42:57Z

You need to squash this to one commit (there are still two commits the first puts the test in their own file, the second moves them into an existing file).

lib/mpl_toolkits/mplot3d/tests/test_axes3d.py

QuLogic · 2022-11-28T23:52:13Z

Thanks @saranti! Congratulations on your first PR to Matplotlib 🎉 We hope to hear from you again.

github-actions bot reviewed Nov 20, 2022

View reviewed changes

story645 reviewed Nov 20, 2022

View reviewed changes

oscargus reviewed Nov 21, 2022

View reviewed changes

tacaswell reviewed Nov 22, 2022

View reviewed changes

tacaswell added this to the v3.7.0 milestone Nov 22, 2022

saranti force-pushed the test branch from da7451d to 4f93e0c Compare November 23, 2022 04:39

Test that plot results aren't affected by mutating input arrays

7a1df78

saranti force-pushed the test branch from f53442b to 7a1df78 Compare November 26, 2022 07:44

QuLogic linked an issue Nov 26, 2022 that may be closed by this pull request

Surprising behaviour of mutating input arrays to Axes.plot vs Axes3D.plot #8990

Closed

QuLogic reviewed Nov 26, 2022

View reviewed changes

lib/mpl_toolkits/mplot3d/tests/test_axes3d.py Outdated Show resolved Hide resolved

lib/mpl_toolkits/mplot3d/tests/test_axes3d.py Outdated Show resolved Hide resolved

Test that plot results aren't affected by mutating input arrays

a30a64e

tacaswell approved these changes Nov 28, 2022

View reviewed changes

QuLogic approved these changes Nov 28, 2022

View reviewed changes

QuLogic merged commit f630c28 into matplotlib:main Nov 28, 2022

saranti deleted the test branch November 29, 2022 02:41

		assert str(pl.get_data_3d()) == result1
		assert str(pl.get_data()) == result2

	@check_figures_equal(extensions=["png"])
	def test_plot_scalar(fig_test, fig_ref):

	result1 = (array([1, 2, 3]), array([0., 0., 0.]), array([0., 0., 0.]))
	result1 = ((1, 2, 3),(0., 0., 0.), (0., 0., 0.))

		assert np.allclose(pl.get_data_3d(), result1) is True
		assert np.allclose(pl.get_data(), result2) is True

Uh oh!

Add test for mutating input arrays #8990 #24511

Add test for mutating input arrays #8990 #24511

Uh oh!

Conversation

saranti commented Nov 20, 2022

PR Summary

PR Checklist

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

oscargus commented Nov 20, 2022

Uh oh!

oscargus left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

saranti Nov 23, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tacaswell commented Nov 22, 2022

Uh oh!

tacaswell commented Nov 23, 2022

Uh oh!

Uh oh!

Uh oh!

QuLogic commented Nov 28, 2022

Uh oh!

Uh oh!

saranti Nov 23, 2022 •

edited

Loading