Best practices

Here are best practices for writing marimo notebooks.

Use global variables sparingly. Keep the number of global variables in your program small to avoid name collisions. If you have intermediate variables, encapsulate them in functions or prefix them with an underscore (_tmp = ...) to make them local to a cell.

Use descriptive names. Use descriptive variable names, especially for global variables. This will help you minimize name clashes, and will also result in better code.

Use functions. Encapsulate logic into functions to avoid polluting the global namespace with temporary or intermediate variables, and to avoid code duplication.

Use Python modules. If your notebook gets too long, split complex logic into helper Python modules and import them into your notebook.

Minimize mutations. marimo does not track mutations to objects. Try to only mutate an object in the cell that creates it, or create new objects instead of mutating existing ones.

Example

Don’t split up declarations and mutations over multiple cells. For example, don’t do this:

l = [1, 2, 3]

l.append(new_item())

Instead, do declare and mutate in the same cell:

l = [1, 2, 3]
...
l.append(new_item())

or, if working in multiple cells, declare a new variable based on the old one:

l = [1, 2, 3]

extended_list = l + [new_item()]

Write idempotent cells. Write cells whose outputs and behavior are the same when given the same inputs (references); such cells are called idempotent. This will help you avoid bugs and cache expensive intermediate computations.

Cache computations with @functools.cache. Use Python’s builtin functools library to cache expensive computations.

For example,

import functools

@functools.cache
def compute_predictions(problem_parameters):
	...

Whenever compute_predictions is called with a value of problem_parameters it has not seen, it will compute the predictions and store them in a cache. The next time it is called with the same parameters, instead of recomputing the predictions, it will return the previously computed value from the cache.