Not just for education and collaboration

Thank you for this article. I already use Jupyter occasionally (and used IPython Notebook earlier), but it taught me many new things that can be done.

The article emphasizes education and collaboration, and while those are surely important, I've found Jupyter convenient for other reasons: it allows experimenting on large data sets without having to wait for them to load every time you start your script. For example RDF files loaded with rdflib tend to be like this. Loading a non-trivial RDF file can take several minutes. If you write an old-fashioned Python script to process a large file, it's frustrating to wait for it to load the data only to see it crash because of a trivial bug after loading. With Jupyter I can approach the problem one step at a time, checking intermediate results and preserving the already loaded data structures. This is of course also possible with the Python REPL, but then it's much more difficult to keep track of what you've done and save the code into a script for later use. With Jupyter, once I'm confident that the process works, I can easily turn the notebook into a regular Python script.