[mbrzusto]

Notebooks where all the rage in the 90's. I spent quite a few years (~15) with Mathematica and its "notebook". Of course, Maple, Mathcad and Matlab all had the same thing. At some point, I wanted to write more readable and modular code that others could use (and that I could reuse) so I switched to using and IDE for python, Java and C. My workflow enables exploratory data analysis as well as algorithm development. The huge advantage of IDE's vs. REPL's is that in the end, it's easier to write the quality of code that is ready to be shipped off to a production server. I have watched the "Return of the Notebook" with some trepidation: yes, it allows quick iteration and lower barrier to entry, but ultimately if your goal is to create a software product, learning good coding skills in an IDE is a much better path.

[takluyver]

> ultimately if your goal is to create a software product, learning good coding skills in an IDE is a much better path

IPython developer here. A lot of the use cases for notebooks are where you're not creating a software product. It's useful where the product you're really after is a scientific result, some plots, or the like, along with a description of how you got that. And where the product is a presentation, a demo, or documentation.

Of course, as you're doing that kind of thing, code that starts out in a notebook often becomes something you want to reuse, and you therefore move it out to an importable module. We are interested in ways to make that process more fluid.

[pgroth]

This is one thing I struggle with. I usually start in a notebook and then eventually want/need to port to a module environment. It would be nice if there were some guidelines or support for this sort of transition. Any pointers?