{
  "id": 668244,
  "title": "installing connected-components-3d for stage_0",
  "url": "/competitions/physionet-ecg-image-digitization/discussion/668244",
  "author_name": "",
  "post_date": "2026-01-15T21:01:28.500781Z",
  "votes": 2,
  "comment_count": 3,
  "views": 0,
  "content": "<p>It looks like Kaggle recently updated the base notebook image from Python 3.11 to Python 3.12. As a result, previously working wheels (e.g., connected-components-3d built for cp311) no longer install due to ABI incompatibility. Has anyone found a reliable way to handle this on Kaggle (e.g., forcing Python 3.11, rebuilding wheels on Kaggle, or alternative packages)?</p>",
  "messages": [
    {
      "id": "3391874",
      "postDate": "01/15/2026 21:01:28",
      "content": "<p>It looks like Kaggle recently updated the base notebook image from Python 3.11 to Python 3.12. As a result, previously working wheels (e.g., connected-components-3d built for cp311) no longer install due to ABI incompatibility. Has anyone found a reliable way to handle this on Kaggle (e.g., forcing Python 3.11, rebuilding wheels on Kaggle, or alternative packages)?</p>",
      "rawMarkdown": "It looks like Kaggle recently updated the base notebook image from Python 3.11 to Python 3.12. As a result, previously working wheels (e.g., connected-components-3d built for cp311) no longer install due to ABI incompatibility. Has anyone found a reliable way to handle this on Kaggle (e.g., forcing Python 3.11, rebuilding wheels on Kaggle, or alternative packages)?",
      "votes": null
    },
    {
      "id": "3391886",
      "postDate": "01/15/2026 21:50:22",
      "content": "<p>Based on my experience, if you choose a public notebook that already installed the connected-components-3d package and then <strong>copy&amp;edit notebook</strong> , you can install the package without much trouble. This approach can temporarily fix the version problem. If anyone knows the real root-cause solution, feel free to share.</p>",
      "rawMarkdown": "Based on my experience, if you choose a public notebook that already installed the connected-components-3d package and then **copy&edit notebook** , you can install the package without much trouble. This approach can temporarily fix the version problem. If anyone knows the real root-cause solution, feel free to share.",
      "votes": null
    },
    {
      "id": "3391888",
      "postDate": "01/15/2026 21:57:27",
      "content": "<p>Thank you for the response, I am doing exactly what you suggested but getting the following, </p>\n<p>!pip uninstall -y tensorflow\n!uv pip install --no-deps --system --no-index --find-links='/kaggle/input/hengck23-submit-physionet/hengck23-submit-physionet/setup' 'connected-components-3d'</p>\n<p>Found existing installation: tensorflow 2.19.0\nUninstalling tensorflow-2.19.0:\n  Successfully uninstalled tensorflow-2.19.0\nUsing Python 3.12.12 environment at: /usr\n  × No solution found when resolving dependencies:\n  ╰─▶ Because only connected-components-3d==3.26.1 is available and\n      connected-components-3d==3.26.1 has no wheels with a matching\n      Python ABI tag (e.g., <code>cp312</code>), we can conclude that all versions of\n      connected-components-3d cannot be used.\n      And because you require connected-components-3d, we can conclude that\n      your requirements are unsatisfiable.</p>\n<pre><code>  hint: You require CPython 3.12 (`cp312`), but we only found wheels for\n  `connected-components-3d` (v3.26.1) with the following Python ABI tag:\n  `cp311`\n</code></pre>",
      "rawMarkdown": "Thank you for the response, I am doing exactly what you suggested but getting the following, \n\n\n!pip uninstall -y tensorflow\n!uv pip install --no-deps --system --no-index --find-links='/kaggle/input/hengck23-submit-physionet/hengck23-submit-physionet/setup' 'connected-components-3d'\n\n\nFound existing installation: tensorflow 2.19.0\nUninstalling tensorflow-2.19.0:\n  Successfully uninstalled tensorflow-2.19.0\nUsing Python 3.12.12 environment at: /usr\n  × No solution found when resolving dependencies:\n  ╰─▶ Because only connected-components-3d==3.26.1 is available and\n      connected-components-3d==3.26.1 has no wheels with a matching\n      Python ABI tag (e.g., `cp312`), we can conclude that all versions of\n      connected-components-3d cannot be used.\n      And because you require connected-components-3d, we can conclude that\n      your requirements are unsatisfiable.\n\n      hint: You require CPython 3.12 (`cp312`), but we only found wheels for\n      `connected-components-3d` (v3.26.1) with the following Python ABI tag:\n      `cp311`",
      "votes": null
    },
    {
      "id": "3391894",
      "postDate": "01/15/2026 22:16:19",
      "content": "<p><a href=\"https://www.kaggle.com/vandongtran\" target=\"_blank\">@vandongtran</a> </p>",
      "rawMarkdown": "vandongtran",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3391886,
      "author_name": "clyde7470",
      "author_url": "",
      "post_date": "01/15/2026 21:50:22",
      "content": "<p>Based on my experience, if you choose a public notebook that already installed the connected-components-3d package and then <strong>copy&amp;edit notebook</strong> , you can install the package without much trouble. This approach can temporarily fix the version problem. If anyone knows the real root-cause solution, feel free to share.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3391888,
          "author_name": "tomi85",
          "author_url": "",
          "post_date": "01/15/2026 21:57:27",
          "content": "<p>Thank you for the response, I am doing exactly what you suggested but getting the following, </p>\n<p>!pip uninstall -y tensorflow\n!uv pip install --no-deps --system --no-index --find-links='/kaggle/input/hengck23-submit-physionet/hengck23-submit-physionet/setup' 'connected-components-3d'</p>\n<p>Found existing installation: tensorflow 2.19.0\nUninstalling tensorflow-2.19.0:\n  Successfully uninstalled tensorflow-2.19.0\nUsing Python 3.12.12 environment at: /usr\n  × No solution found when resolving dependencies:\n  ╰─▶ Because only connected-components-3d==3.26.1 is available and\n      connected-components-3d==3.26.1 has no wheels with a matching\n      Python ABI tag (e.g., <code>cp312</code>), we can conclude that all versions of\n      connected-components-3d cannot be used.\n      And because you require connected-components-3d, we can conclude that\n      your requirements are unsatisfiable.</p>\n<pre><code>  hint: You require CPython 3.12 (`cp312`), but we only found wheels for\n  `connected-components-3d` (v3.26.1) with the following Python ABI tag:\n  `cp311`\n</code></pre>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 3391894,
      "author_name": "banhmimatong",
      "author_url": "",
      "post_date": "01/15/2026 22:16:19",
      "content": "<p><a href=\"https://www.kaggle.com/vandongtran\" target=\"_blank\">@vandongtran</a> </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3391874": "It looks like Kaggle recently updated the base notebook image from Python 3.11 to Python 3.12. As a result, previously working wheels (e.g., connected-components-3d built for cp311) no longer install due to ABI incompatibility. Has anyone found a reliable way to handle this on Kaggle (e.g., forcing Python 3.11, rebuilding wheels on Kaggle, or alternative packages)?",
    "3391886": "Based on my experience, if you choose a public notebook that already installed the connected-components-3d package and then **copy&edit notebook** , you can install the package without much trouble. This approach can temporarily fix the version problem. If anyone knows the real root-cause solution, feel free to share.",
    "3391888": "Thank you for the response, I am doing exactly what you suggested but getting the following, \n\n\n!pip uninstall -y tensorflow\n!uv pip install --no-deps --system --no-index --find-links='/kaggle/input/hengck23-submit-physionet/hengck23-submit-physionet/setup' 'connected-components-3d'\n\n\nFound existing installation: tensorflow 2.19.0\nUninstalling tensorflow-2.19.0:\n  Successfully uninstalled tensorflow-2.19.0\nUsing Python 3.12.12 environment at: /usr\n  × No solution found when resolving dependencies:\n  ╰─▶ Because only connected-components-3d==3.26.1 is available and\n      connected-components-3d==3.26.1 has no wheels with a matching\n      Python ABI tag (e.g., `cp312`), we can conclude that all versions of\n      connected-components-3d cannot be used.\n      And because you require connected-components-3d, we can conclude that\n      your requirements are unsatisfiable.\n\n      hint: You require CPython 3.12 (`cp312`), but we only found wheels for\n      `connected-components-3d` (v3.26.1) with the following Python ABI tag:\n      `cp311`",
    "3391894": "vandongtran"
  },
  "source": "meta"
}