René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"[Bug]: pcolormesh writing to input mask","articleBody":"### Bug summary\n\nWhen `pcolormesh` receives a masked array, it seems to be writing back to the mask.  Since numpy 1.24 this now causes `pcolormesh` to fail if the mask is read-only.\n\n### Code for reproduction\n\n```python\nimport matplotlib.pyplot as plt\r\nimport numpy as np\r\n\r\ndata = np.arange(6).reshape(2, 3)\r\nmask = np.broadcast_to([False, True, False], data.shape)  # read-only array\r\n\r\nmasked_data = np.ma.array(data, mask=mask)\r\n\r\nplt.pcolormesh(masked_data)\n```\n\n\n### Actual outcome\n\n```\r\nTraceback (most recent call last):\r\n  File \"pcolormesh_read_only_mask.py\", line 9, in \u003cmodule\u003e\r\n    plt.pcolormesh(masked_data)\r\n  File \"[conda-env-path]/lib/python3.11/site-packages/matplotlib/pyplot.py\", line 2773, in pcolormesh\r\n    __ret = gca().pcolormesh(\r\n            ^^^^^^^^^^^^^^^^^\r\n  File \"[conda-env-path]/lib/python3.11/site-packages/matplotlib/__init__.py\", line 1442, in inner\r\n    return func(ax, *map(sanitize_sequence, args), **kwargs)\r\n           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n  File \"[conda-env-path]/lib/python3.11/site-packages/matplotlib/axes/_axes.py\", line 6220, in pcolormesh\r\n    X, Y, C, shading = self._pcolorargs('pcolormesh', *args,\r\n                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n  File \"[conda-env-path]/lib/python3.11/site-packages/matplotlib/axes/_axes.py\", line 5704, in _pcolorargs\r\n    C = cbook.safe_masked_invalid(C)\r\n        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n  File \"[conda-env-path]/lib/python3.11/site-packages/matplotlib/cbook/__init__.py\", line 715, in safe_masked_invalid\r\n    xm = np.ma.masked_invalid(x, copy=False)\r\n         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n  File \"[conda-env-path]/lib/python3.11/site-packages/numpy/ma/core.py\", line 2360, in masked_invalid\r\n    res = masked_where(~(np.isfinite(a)), a, copy=copy)\r\n          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n  File \"[conda-env-path]/lib/python3.11/site-packages/numpy/ma/core.py\", line 1942, in masked_where\r\n    result.mask = _shrink_mask(cond)\r\n    ^^^^^^^^^^^\r\n  File \"[conda-env-path]/lib/python3.11/site-packages/numpy/ma/core.py\", line 3516, in mask\r\n    self.__setmask__(value)\r\n  File \"[conda-env-path]/lib/python3.11/site-packages/numpy/ma/core.py\", line 3462, in __setmask__\r\n    current_mask.flat = mask\r\n    ^^^^^^^^^^^^^^^^^\r\nValueError: array is read-only\r\n```\n\n### Expected outcome\n\nNo error\n\n### Additional information\n\nThe code above runs fine with numpy v1.23, although the output from `broadcast_to` was already read-only at that version.  From numpy release notes, this looks like the likely reason for the change:\r\nhttps://numpy.org/doc/stable/release/1.24.0-notes.html#masked-invalid-now-modifies-the-mask-in-place\r\n\r\nAside from the new error, if a user passes a masked array that has nans or infs at the unmasked points, we are modifying their input array with the call to `masked_invalid`.\r\n\r\nI guess we just need to take a copy somewhere?\n\n### Operating system\n\nRHEL7\n\n### Matplotlib Version\n\n3.7.1\n\n### Matplotlib Backend\n\nQtAgg\n\n### Python version\n\n3.11.3\n\n### Jupyter version\n\nN/A\n\n### Installation\n\nconda","author":{"url":"https://github.com/rcomer","@type":"Person","name":"rcomer"},"datePublished":"2023-06-08T09:28:32.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":2},"url":"https://github.com/26093/matplotlib/issues/26093"}