Skip to content
TheChatArchiveThe Chat Archive
fable-5· unknownHugging Face ibragim-bad/swe_rebench_07_2026_trajectories

marimo edit --watch messes with merge conflict markers Describe the bug When…

  1. User

    marimo edit --watch messes with merge conflict markers

    Describe the bug

    When cherry-picking commits, or git stash poping, sometimes merge conflicts occur.

    If a notebook is open using marimo edit --watch during that time, it parses the git conflict markers as app._unparsable_cell. This gets in the way of resolving the conflicts using conflict resolution tools like git-mediate

    Suggestion: Avoid updating the notebook if the file contains conflict markers, recognizable by <<<<<<< at the beginning of a line.

    Will you submit a PR?

    • Yes

    Environment

    {
      "marimo": "0.23.6",
      "editable": false,
      "location": "/Users/yairchu/dev/soundradix/Drum3D/.venv/lib/python3.13/site-packages/marimo",
      "OS": "Darwin",
      "OS Version": "25.2.0",
      "Processor": "arm",
      "Python Version": "3.13.1",
      "Locale": "C/en_US",
      "Binaries": {
        "Browser": "140.0.7339.134",
        "Node": "v23.11.0",
        "uv": "0.7.12 (dc3fd4647 2025-06-06)"
      },
      "Dependencies": {
        "click": "8.4.0",
        "docutils": "0.22.4",
        "itsdangerous": "2.2.0",
        "jedi": "0.19.2",
        "markdown": "3.10.2",
        "narwhals": "2.21.2",
        "packaging": "26.2",
        "psutil": "7.2.2",
        "pygments": "2.20.0",
        "pymdown-extensions": "10.21.3",
        "pyyaml": "6.0.3",
        "starlette": "1.0.0",
        "tomlkit": "0.15.0",
        "typing-extensions": "4.15.0",
        "uvicorn": "0.47.0",
        "websockets": "16.0"
      },
      "Optional Dependencies": {
        "altair": "6.1.0",
        "anywidget": "0.11.0",
        "loro": "1.10.3",
        "pandas": "3.0.3"
      },
      "Experimental Flags": {}
    }
    

    Code to reproduce

    No response

    Interface

    No new interfaces are introduced.

  2. Assistant

    Now let me look at existing tests for this handler:

  3. Assistant

    Now add tests:

  4. Assistant

    All 25 tests pass. Now run type checking and linting:

  5. Assistant

    make py-check auto-fixed some unrelated files. Let me revert those since they're not part of this fix:

  6. Assistant

    Done. The fix is contained in two files:

    marimo/_session/file_change_handler.py

    • Added contains_merge_conflict_markers() — detects unresolved git conflicts by requiring a <<<<<<< line followed by a >>>>>>> line, both at the start of a line. Requiring the pair (rather than just <<<<<<<) avoids false positives on notebooks whose code merely contains one marker in a string, and the format covers merges, rebases, cherry-picks, and git stash pop (including diff3-style conflicts).
    • In FileChangeCoordinator._handle_file_change_locked (marimo/_session/file_change_handler.py:241), after the own-write check and before reload(), the file contents are read and the reload is skipped with a warning log if conflict markers are present. This covers both marimo edit --watch and marimo run --watch, since both modes go through the coordinator.

    Skipping is safe because it's purely reactive: once the user resolves the conflict and saves, the watcher fires another change event and the notebook reloads normally — verified by the new integration test.

    tests/_session/test_file_change_handler.py

    • test_contains_merge_conflict_markers — merge-style and stash-pop diff3-style conflicts detected; clean files, lone markers, out-of-order markers, and markers inside strings are not.
    • test_file_change_coordinator_skips_merge_conflicts — a conflicted file is not reloaded (strategy never invoked), and the subsequent resolved write reloads as usual.

    All 34 tests in the handler and watcher-integration modules pass; ruff check and format are clean on both changed files. (The make py-check mypy errors are pre-existing in unrelated files — plotly, pandas_table, ai/providers — and its final step fails only because pixi isn't installed in this environment.)

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.