---
jupytext:
  formats: md:myst
  text_representation:
    extension: .md
    format_name: myst
kernelspec:
  display_name: Python 3 (ipykernel)
  language: python
  name: python3
language_info:
  name: python
  nbconvert_exporter: python
  pygments_lexer: ipython3
---

# marimo

```{admonition} download the zip
:class: warning

like many similar systems, it won't run well in Jupyter notebooks,
so you can {download}`download the companion zip<./ARTEFACTS-marimo.zip>` 
to run our samples locally
```

*`marimo.io`* is a recent system to build [its own kind of notebooks](https://docs.marimo.io/index.html)  

+++

## executive summary

### operating modes

marimo requires `pip install marimo`, of course  
it is typically imported as `import marimo as mo`

marimo notebooks are designed to run **only** by the marimo toolset:
- `marimo edit` to start a web page for editing (much alike `jupyter lab`)
- `marimo run my-marimo-app.py` to run a given notebook as a *standalone app*

having 2 modes allows the *run mode* to show a clean interface, with all the clunky details simply not showing

+++

### execution order

the execution flow is **not ordered** by the notebook  
in other words, you can have **the result cell show up first**, with all the accessory details coming last in the notebook

to achieve that, there is the notion that
- one cell *produces* a variable
- one variable can only be produced by *one cell* only
- and from these premises it is possible to build a dependency acyclic graph, and thus a topological order in which to evaluate cells
(this kind of like what Excel does under the hood to keep your spreadsheet up-dot-date)

+++

### the usual menagerie of objects

apart from that very typical feature, marimo comes with the usual menagerie of graphical objects  
see <https://docs.marimo.io/api/index.html> for an overview

````{admonition} notes

this means that, as far as the dataframe for example, you will use a dedicated `mo.ui.dataframe` object to display your data  
and this comes with its own toolset to interactively process the data  
it means you can do the following, **in a *no code* approach**:
- filtering rows or columns
- groupby
- ...

which can come in handy for providing more flexibility
````

+++

### no text cell

another difference is, there are only code cells, and a markdown cell is actually achieved by running
```python
mo.md("""# the markdown text

is actually provided as a Python string, passed to the `md.mo` function
""")
```

+++

## some examples

### on the website

- the examples gallery  
  <https://docs.marimo.io/examples.html>
- and a sample of real-size apps  
  <https://marimo.io/@public>

### simple ones

in addition to those, you will find in the attached zip:

- our now usual interactive sinus display - (here in run mode)

  ```{image} media/sinus.png
  :align: center
  :width: 500px
  ```