Closer to production quality Python notebooks with `marimo check`

Post by sixo » Thu, Oct 16, 2025, 7:15 PM UTC

dmadisetti wrote: post #13

sixo wrote: post #10

Earlier quoted context omitted.

Unfortunately the code in Marimo notebooks implements cross-variable references as untyped function arguments, meaning that nothing passed between cells can be type-safe. This makes it very hard to use tooling directly on the Python representation of notebooks.

Arguments are typed if explicitly specified in definition, e.g. ```python @app.cell def my_cell(): x : int = 1 return (x,) @app.cell def consumer(x : int): y = x return ``` We've talked about building out implicit typing into our serialization- but that may be a larger can of worms.

    import numpy

    @app.cell
    def my_cell(np: types.ModuleType[numpy]):
        ... editor treats np like `import numpy as np` ...

Post by sixo » Sun, Oct 19, 2025, 1:11 AM UTC

dmadisetti wrote: post #22

sixo wrote: post #21

Earlier quoted context omitted.

That would be wonderful. A lot of the arguments are just imports, and these would have probably have the largest upside. It would be great if there was a construct like import numpy @app.cell def my_cell(np: types.ModuleType[numpy]): ... editor treats np like `import numpy as np` ... I do not use Python enough to know if something like this can be hacked together. But if not, I imagine could be PIPed into the standar…

We've addressed this too by allowing a setup cell: ... with app.setup: import numpy as np @app.cell def cell_uses_np(): # no np needed in signature result = np.... The best part about this pattern is that it enables "pure" (only dependent on setup cell) functions to be addressable: with app.setup: from typing import Optional, Any @app.function def my_fn(arg: Optional[Any]) -> Any: ... but nice convergent design :) gl…