René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"[Bug]: Intermittent `ValueError: The passed figure is not managed by pyplot` with ipympl (`%matplotlib widget`)","articleBody":"### Bug summary\n\nWhen using Matplotlib with ipympl (`%matplotlib widget`) in JupyterLab, I intermittently get:\n\n```\nValueError: The passed figure is not managed by pyplot\n```\n\nThis happens with standard pyplot usage (`plt.figure`, `plt.subplots(num=fig)`, `plt.sca(ax)`), without explicitly constructing canvases.\n\n### Code for reproduction\n\n```Python\n%matplotlib widget\nimport time\nimport matplotlib.pyplot as plt\n\ndef try_once():\n    fig_num = 1\n    plt.close(fig_num)\n    fig = plt.figure(fig_num)\n\n    fig, axs = plt.subplots(10, 10, num=fig, squeeze=False)\n\n    for ax in axs.flat:\n        plt.sca(ax)  # \u003c-- intermittently raises\n        plt.plot([1, 2, 3])\n        fig.canvas.draw_idle()\n        time.sleep(0.001)\n\nfor i in range(300):\n    try:\n        try_once()\n    except Exception as e:\n        print('FAILED at iteration', i, '-\u003e', repr(e))\n        raise\n```\n\n### Actual outcome\n\n![Image](https://github.com/user-attachments/assets/fd27ca1f-0d43-40c7-b8f4-cfcd474c9a1e)\n\n```\n---------------------------------------------------------------------------\nValueError                                Traceback (most recent call last)\nCell In[2], line 20\n     18 for i in range(300):\n     19     try:\n---\u003e 20         try_once()\n     21     except Exception as e:\n     22         print('FAILED at iteration', i, '-\u003e', repr(e))\n\nCell In[2], line 13, in try_once()\n     10 fig, axs = plt.subplots(10, 10, num=fig, squeeze=False)\n     12 for ax in axs.flat:\n---\u003e 13     plt.sca(ax)  # \u003c-- intermittently raises\n     14     plt.plot([1, 2, 3])\n     15     fig.canvas.draw_idle()\n\nFile ~/ipympl-debug/issue/.venv/lib/python3.12/site-packages/matplotlib/pyplot.py:1376, in sca(ax)\n   1372 # Mypy sees ax.figure as potentially None,\n   1373 # but if you are calling this, it won't be None\n   1374 # Additionally the slight difference between `Figure` and `FigureBase` mypy catches\n   1375 fig = ax.get_figure(root=False)\n-\u003e 1376 figure(fig)  # type: ignore[arg-type]\n   1377 fig.sca(ax)\n\nFile ~/ipympl-debug/issue/.venv/lib/python3.12/site-packages/matplotlib/pyplot.py:995, in figure(num, figsize, dpi, facecolor, edgecolor, frameon, FigureClass, clear, **kwargs)\n    993 root_fig = num.get_figure(root=True)\n    994 if root_fig.canvas.manager is None:\n--\u003e 995     raise ValueError(\"The passed figure is not managed by pyplot\")\n    996 elif (any(param is not None for param in [figsize, dpi, facecolor, edgecolor])\n    997       or not frameon or kwargs) and root_fig.canvas.manager.num in allnums:\n    998     _api.warn_external(\n    999         \"Ignoring specified arguments in this call because figure \"\n   1000         f\"with num: {root_fig.canvas.manager.num} already exists\")\n\nValueError: The passed figure is not managed by pyplot\n```\n\n### Expected outcome\n\nMosaicked plots will be shown without errors.\n\n### Additional information\n\n### Observed state at failure\n\nAt the moment the exception occurs:\n- `root_fig.canvas.manager is None`\n- but the corresponding `FigureManager` still exists in `matplotlib._pylab_helpers.Gcf` (so pyplot still manages the figure)\n\nSo pyplot’s “managed figure” check is making a false negative.\n\n### Where the exception comes from\n\n`plt.sca(ax)` calls `pyplot.figure(fig)` internally.\nWhen passed a `Figure`, `pyplot.figure()` currently checks only:\n\n```python\nroot_fig = num.get_figure(root=True)\nif root_fig.canvas.manager is None:\n    raise ValueError(\"The passed figure is not managed by pyplot\")\n```\n\nIn this bug, `canvas.manager` is temporarily/incorrectly `None` even though the manager still exists in `Gcf`.\n\n### Proposed fix\n\nIf `root_fig.canvas.manager is None`, fall back to a `Gcf` lookup to find the manager whose `m.canvas.figure is root_fig`, and reattach it.\n\n```python\nmanager = root_fig.canvas.manager\nif manager is None:\n    for m in _pylab_helpers.Gcf.get_all_fig_managers():\n        if getattr(getattr(m, \"canvas\", None), \"figure\", None) is root_fig:\n            manager = m\n            root_fig.canvas.manager = m\n            break\nif manager is None:\n    raise ValueError(...)\n```\n\nThis same logic eliminates the failures in my environment (0/60) when applied as a small monkey patch around `pyplot.figure()`.\n\nThis resembles matplotlib/matplotlib#19380 (same structural cause: `FigureCanvasBase.__init__` sets `self.manager=None` after replacing `figure.canvas`).\n\n### Operating system\n\nUbuntu 24\n\n### Matplotlib Version\n\n3.10.8\n\n### ipympl Version\n\n0.10.0\n\n### Matplotlib Backend\n\nwidget\n\n### Python version\n\nPython 3.12.3\n\n### Jupyter version\n\njupyter lab --version: 4.5.4\n\n### Installation\n\nuv","author":{"url":"https://github.com/michitaro","@type":"Person","name":"michitaro"},"datePublished":"2026-03-02T05:29:45.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":13},"url":"https://github.com/622/ipympl/issues/622"}