TST: Use correct method of clearing mock objects by QuLogic · Pull Request #31153 · matplotlib/matplotlib · GitHub

PR summary

In Python 3.13.12 and 3.14.3, fixes for thread-safety of mock call_count meant that manually changing it no longer works [1]. Instead use the more correct reset_mock method.

[1] python/cpython#142651 (comment)

Fixes #31112

PR checklist

"closes #0000" is in the body of the PR description to link the related issue
new and changed code is tested
[n/a] Plotting related features are demonstrated in an example
[n/a] New Features and API Changes are noted with a directive and release note
[n/a] Documentation complies with general and docstring guidelines

In Python 3.13.12 and 3.14.3, fixes for thread-safety of mock `call_count` meant that manually changing it no longer works [1]. Instead use the more correct `reset_mock` method. [1] python/cpython#142651 (comment)

[gw0] linux -- Python 3.11.14 /opt/hostedtoolcache/Python/3.11.14/x64/bin/python
173

174
fmt = '%4.2e'
175

176
    @pytest.mark.parametrize('fmt', ['%4.2e', '{x:.2e}'])
177
    def test_colorbar_format(fmt):
178
        # make sure that format is passed properly
179
        x, y = np.ogrid[-4:4:31j, -4:4:31j]
180
        z = 120000*np.exp(-x**2 - y**2)
181
    
182
        fig, ax = plt.subplots()
183
        im = ax.imshow(z)
184
        cbar = fig.colorbar(im, format=fmt)
185
        fig.canvas.draw()
186
        assert cbar.ax.yaxis.get_ticklabels()[4].get_text() == '8.00e+04'
187
    
188
        # make sure that if we change the clim of the mappable that the
189
        # formatting is *not* lost:
190
        im.set_clim([4, 200])
191
        fig.canvas.draw()
192
        assert cbar.ax.yaxis.get_ticklabels()[4].get_text() == '2.00e+02'
193
    
194
        # but if we change the norm:
195
        im.set_norm(LogNorm(vmin=0.1, vmax=10))
196
        fig.canvas.draw()
197
>       assert (cbar.ax.yaxis.get_ticklabels()[0].get_text() ==
198
                '$\\mathdefault{10^{-2}}$')
199
E       AssertionError: assert '$\\mathdefault{10^{-1}}$' == '$\\mathdefault{10^{-2}}$'
200
E         - $\mathdefault{10^{-2}}$
201
E         ?                    ^
202
E         + $\mathdefault{10^{-1}}$
203
E         ?                    ^
204

205
lib/matplotlib/tests/test_colorbar.py:637: AssertionError

That is a failure I have never seen before.

Restarted the job to see if it is transient.

I‘ve seen this error somewhere before, but don’t remember where.

Tests passed on re-run, something to keep an eye out for if we see it more, but treating as a one-off until then, and this is a patch which fixes our tests, so merging now

…k objects

…tation means it can no longer just be set to 0 (#10576) This should fix our failing tests. Python 3.14.3 has the following in its changelog: > [gh-142651](python/cpython#142651): [unittest.mock](https://docs.python.org/3/library/unittest.mock.html#module-unittest.mock): fix a thread safety issue where [Mock.call_count](https://docs.python.org/3/library/unittest.mock.html#unittest.mock.Mock.call_count) may return inaccurate values when the mock is called concurrently from multiple threads. As a result, we have to call `reset_mock()` instead of using `.call_count = 0`. See example [here](matplotlib/matplotlib#31153). Tests on my machine showing that this change fixes things, and it's the only place to fix: ```bash $ brew upgrade pyenv $ pyenv install 3.14.3 $ pyenv global 3.14.3 $ tools/venv.py $ source venv/bin/activate $ pytest certbot -k "test_rollback_too_many" ====================================================================== test session starts ======================================================================= platform darwin -- Python 3.14.3, pytest-9.0.2, pluggy-1.6.0 rootdir: /Users/erica/certbot configfile: pytest.ini plugins: anyio-4.12.1, xdist-3.8.0, cov-7.0.0 collected 1039 items / 1038 deselected / 1 selected certbot/src/certbot/_internal/tests/reverter_test.py . [100%] =============================================================== 1 passed, 1038 deselected in 2.94s =============================================================== $ git grep 'call_count = 0' $ git checkout main $ pytest certbot -k "test_rollback_too_many" ====================================================================== test session starts ======================================================================= platform darwin -- Python 3.14.3, pytest-9.0.2, pluggy-1.6.0 rootdir: /Users/erica/certbot configfile: pytest.ini plugins: anyio-4.12.1, xdist-3.8.0, cov-7.0.0 collected 1039 items / 1038 deselected / 1 selected certbot/src/certbot/_internal/tests/reverter_test.py F [100%] ============================================================================ FAILURES ============================================================================ _______________________________________________________ TestFullCheckpointsReverter.test_rollback_too_many _______________________________________________________ self = <certbot._internal.tests.reverter_test.TestFullCheckpointsReverter testMethod=test_rollback_too_many> mock_logger = <MagicMock name='logger' id='4463351456'> > ??? E AssertionError: assert 2 == 1 E + where 2 = <MagicMock name='logger.warning' id='4463351792'>.call_count E + where <MagicMock name='logger.warning' id='4463351792'> = <MagicMock name='logger' id='4463351456'>.warning certbot/src/certbot/_internal/tests/reverter_test.py:363: AssertionError ==================================================================== short test summary info ===================================================================== FAILED certbot/src/certbot/_internal/tests/reverter_test.py::TestFullCheckpointsReverter::test_rollback_too_many - AssertionError: assert 2 == 1 =============================================================== 1 failed, 1038 deselected in 0.48s =============================================================== $ git grep 'call_count = 0' certbot/src/certbot/_internal/tests/reverter_test.py: mock_logger.warning.call_count = 0 ```

…153-on-v3.10.x Backport PR #31153 on branch v3.10.x (TST: Use correct method of clearing mock objects)

https://build.opensuse.org/request/show/1333726 by user dgarcia + anag_factory - TST: Use correct method of clearing mock objects - Upstream patch correct-reset-mock.patch, gh#matplotlib/matplotlib#31153

…k objects

…t-of-pr-31153-on-v3.10.x Backport PR matplotlib#31153 on branch v3.10.x (TST: Use correct method of clearing mock objects)

TST: Use correct method of clearing mock objects

63207bb

In Python 3.13.12 and 3.14.3, fixes for thread-safety of mock `call_count` meant that manually changing it no longer works [1]. Instead use the more correct `reset_mock` method. [1] python/cpython#142651 (comment)

QuLogic added this to the v3.10.9 milestone Feb 13, 2026

QuLogic added the topic: testing label Feb 13, 2026

ohemorange mentioned this pull request Feb 13, 2026

Reset mock call count using reset_mock since new thread-safe implementation means it can no longer just be set to 0 certbot/certbot#10576

Merged

greglucas approved these changes Feb 13, 2026

View reviewed changes

QuLogic mentioned this pull request Feb 13, 2026

ci: Try pinning an older Python #31152

Closed

2 tasks

timhoffm approved these changes Feb 13, 2026

View reviewed changes

tacaswell approved these changes Feb 13, 2026

View reviewed changes

ksunden merged commit d3b539c into matplotlib:main Feb 13, 2026
54 of 55 checks passed

meeseeksmachine pushed a commit to meeseeksmachine/matplotlib that referenced this pull request Feb 13, 2026

Backport PR matplotlib#31153 : TST: Use correct method of clearing moc…

b9d7e7f

…k objects

meeseeksmachine mentioned this pull request Feb 13, 2026

Backport PR #31153 on branch v3.10.x (TST: Use correct method of clearing mock objects) #31154

Merged

QuLogic deleted the fix-mock-usage branch February 13, 2026 19:23

timhoffm added a commit that referenced this pull request Feb 13, 2026

Merge pull request #31154 from meeseeksmachine/auto-backport-of-pr-31…

cc7fe5a

…153-on-v3.10.x Backport PR #31153 on branch v3.10.x (TST: Use correct method of clearing mock objects)

wavebyrd pushed a commit to wavebyrd/matplotlib that referenced this pull request Mar 13, 2026

Backport PR matplotlib#31153 : TST: Use correct method of clearing moc…

7ebe6c2

…k objects

QuLogic mentioned this pull request Mar 31, 2026

Incompatibility with Python 3.15 alpha 7 #31429

Closed

This was referenced Apr 21, 2026

Backport PR #31282 on branch v3.10.x (SEC: Block shell escapes in latex and ps commands) #31542

Closed

Backport PR #31248 on branch v3.10.x (SEC: Remove eval() from validate_cycler) #31543

Closed

merge up v3.10.9 #31560

Merged

TST: Use correct method of clearing mock objects by QuLogic · Pull Request #31153 · matplotlib/matplotlib · GitHub

Uh oh!

Uh oh!

TST: Use correct method of clearing mock objects - #31153

TST: Use correct method of clearing mock objects#31153
ksunden merged 1 commit into
matplotlib:mainfrom
QuLogic:fix-mock-usage

QuLogic commented Feb 13, 2026

Uh oh!

tacaswell commented Feb 13, 2026

Uh oh!

tacaswell commented Feb 13, 2026

Uh oh!

timhoffm commented Feb 13, 2026

Uh oh!

ksunden commented Feb 13, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

TST: Use correct method of clearing mock objects by QuLogic · Pull Request #31153 · matplotlib/matplotlib · GitHub

Uh oh!

Uh oh!

TST: Use correct method of clearing mock objects - #31153

TST: Use correct method of clearing mock objects#31153ksunden merged 1 commit intomatplotlib:mainmatplotlib/matplotlib:mainfrom QuLogic:fix-mock-usageQuLogic/matplotlib:fix-mock-usageCopy head branch name to clipboard

Conversation

QuLogic commented Feb 13, 2026

PR summary

PR checklist

Uh oh!

tacaswell commented Feb 13, 2026

Uh oh!

tacaswell commented Feb 13, 2026

Uh oh!

timhoffm commented Feb 13, 2026

Uh oh!

ksunden commented Feb 13, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

TST: Use correct method of clearing mock objects#31153
ksunden merged 1 commit into
matplotlib:mainfrom
QuLogic:fix-mock-usage