It took me a solid decade from the time I first tried them out until I came around to seeing how notebooks could be useful for me. (For a long time, I thought they were just a development environment with limitations I didn’t care for; eventually I started to see them as a REPL with some superpowers.)
Lately, though, I think I prefer marimo for most notebook-y things I do. It gives a lot of the same REPL-with-superpowers utility, while being more amenable to things like source control, linters, formatters, etc. And it’s easier for me to share with others than Jupyter-based things are. It’s basically what would happen if you mashed the best parts of jupyter and streamlit together, IMO. I’m a little concerned with the funding source, but it’s Apache licensed, and I have a hard time worrying any more about it than I do about uv, ruff, etc.
I’m not sure I could be persuaded to use a Jupyter-specific IDE.
It saves you from having to remember to restart your kernel and rerunning the notebook from the beginning just to know whether the notebook will run the next time you open it.
The “quick start” consists of instructions for building the optional Electron front-end and the Go back-end from the GitHub project source; https://zasper.io/downloads are all “coming soon”. Early days?
I had to help someone non-technical do some basic Python scripting on windows, and getting a dev environment set up on windows when I was only familiar with Linux/Mac was one of the worst experiences I have ever encountered. I had never really appreciated Notebooks before that.
It’s basically an extended REPL, so exploratory programming, and documentation of that. Often “data science” things: load a dataset, run different filters, tests and visualizations interactively against it in a REPL, then have a record of that with all the intermediate results to store, share, and modify as needed.
It took me a solid decade from the time I first tried them out until I came around to seeing how notebooks could be useful for me. (For a long time, I thought they were just a development environment with limitations I didn’t care for; eventually I started to see them as a REPL with some superpowers.)
Lately, though, I think I prefer marimo for most notebook-y things I do. It gives a lot of the same REPL-with-superpowers utility, while being more amenable to things like source control, linters, formatters, etc. And it’s easier for me to share with others than Jupyter-based things are. It’s basically what would happen if you mashed the best parts of jupyter and streamlit together, IMO. I’m a little concerned with the funding source, but it’s Apache licensed, and I have a hard time worrying any more about it than I do about uv, ruff, etc.
I’m not sure I could be persuaded to use a Jupyter-specific IDE.
Wow, marimo looks really slick! I love that the code can run in the browser, I’ve always wanted Jupyter but with more client, less server.
While I like competition the JupyterLab UI never really struck me as slow. Also what makes JupyterLab great is the ecosystem.
(Update: After writing this, I realized it’s basically a dupe of https://lobste.rs/s/tz6ybl/zasper_fastest_most_efficient_ide_run#c_lsbkbz. I’m new here, not sure whether the polite thing is to delete the comment or to keep it. Also, I guess my comment has a few more links)
I’ve really enjoyed the reactivity of https://github.com/fonsp/Pluto.jl. These days, there is an analogous project in the Python world, https://marimo.app/. See also https://lobste.rs/s/dezuac/marimo_next_generation_python_notebook
It saves you from having to remember to restart your kernel and rerunning the notebook from the beginning just to know whether the notebook will run the next time you open it.
The “quick start” consists of instructions for building the optional Electron front-end and the Go back-end from the GitHub project source; https://zasper.io/downloads are all “coming soon”. Early days?
I had to help someone non-technical do some basic Python scripting on windows, and getting a dev environment set up on windows when I was only familiar with Linux/Mac was one of the worst experiences I have ever encountered. I had never really appreciated Notebooks before that.
Can someone explain the use case for these notebooks?
It’s basically an extended REPL, so exploratory programming, and documentation of that. Often “data science” things: load a dataset, run different filters, tests and visualizations interactively against it in a REPL, then have a record of that with all the intermediate results to store, share, and modify as needed.