New post: The Technical Guide to Jupyter Setup (Part 2 of my series).
If you're still doing pip install jupyter and committing JSON blobs to Git, there's a better way.
I'm diving into:
✅ Using #uv for fast envs.
✅ #Jupytext for pairing notebooks with .py files.
✅ Professional project hierarchies.
✅ #VSCode integration for #Copilot users.
Treat your research like production code.
Read more: https://g.omid.dev/FpT0kTO
@adamsteer I feel like #JupyText solves most of the issues people complain about - your notebook is now a .py file you can run as an executable script or open/edit easily in a different text editor should you or someone else wish to.
Two tools called out for automating the process of extracting reusable code from #Jupyter Notebooks:
#nbconvert: https://nbconvert.readthedocs.io/en/latest/
#jupytext: https://jupytext.readthedocs.io/en/latest/
Jupytext has a feature that allows for notebooks and scripts to keep in sync. And both of them might work well for simpler notebooks, definitely try them out to save time.
I'm a bit frustrated that the markdown generated by #jupytext does not contain the images resulting from the cell executions... I can imagine it would be hard or impractical to make it work like that, but I'm not sure how to build a workflow out of it.
Have you considered #Jupytext https://jupytext.readthedocs.io/en/latest?
it allows to create and transform markdown documents from notebooks in #Jupyter, these markdown files can easily be used in version control, and you even can edit with your favourite editor (in my case is Emacs, but you can use any other like vim). Also, it can be automatically sync with the notebooks, so in case your colleagues prefer use notebooks, the can edit them directly, and the markdown is always updated.
#Jupytext is now fully compatible with #JupyterLab 4 and Jupyter Notebook 7. And the Jupytext menu is back! Check out the latest Jupytext news at https://blog.jupyter.org/the-jupytext-menu-is-back-3e6212e8c090, and see how text #JupyterNotebooks can enhance your development/research workflow!
#Jupyter+#Jupytext. I just discovered Jupytext and it's completely changed my life. Suddenly my project which was to be code, documentation, and discourse trying to explain what it was all about has become just notebook chapters that are meant to be read, run, and imported as libraries. That's just terrific. Now my sub-modules are chapters, and my whole project without me realizing it has turned into a "book" in progress - which is actually the best way to present it all.
Using #MarkdownSQLPro Tools 🛠️ with new #SQLBookmarks 🔖, #SQLite & our #DuckDBPro #SQLTools in #VSCode IDE on @ploomber's #JupySQL #JupyText / #MyST #JupyterBook 📓 markdown docs 😎
#SQL #DuckDB #Markdown #ProDataTools 🧙♂️ ...
@philiph have u tried the tool that converts and syncs notebooks to .Md files #jupytext and #nbdime well for us. https://blog.reviewnb.com/tools-for-git-jupyter-workflow/ I haven’t tried #nbdime yet but in a workflow which is quick and dirty then getting more serious jupytext is nice and integrated documents with code overall is nice. Then u can use nbdev to turn it into a module.
If you find yourself tracking revisions of #Jupyter #notebooks on #git (or an inferior #VCS), you will LOVE #Jupytext!
Since the text editor Atom (+Hydrogen) was sunset, I found myself coming back to Jupyter more often for interactive stuff. Now I can save notebooks as .py (Markdown in comments) or as .md (Python in code blocks) -- or both and have one updated when I change the other. And the best thing: Thanks to Jupytext they run and feel exactly as .ipynb. #python #ide
