There are many ways to go about programming. One of the most productive paradigms is interactive: You use a REPL (read-eval-print loop) to write and test your code as you code, and then copy the tested code into a file.
The REPL method, which originated in LISP development environments, is well-suited to Python programming, as Python has always had good interactive development tools. The drawback of this style of programming is that once you’ve written the code you have to separately pull out the tests and write the documentation, save all that to a repository, do your packaging, and publish your package and documentation.
Donald Knuth’s literate programming paradigm prescribes writing the documentation and code in the same document, with the documentation aimed at humans interspersed with the code intended for the computer. Literate programming has been used widely for scientific programming and data science, often using notebook environments, such as Jupyter Notebooks, Jupyter Lab, Visual Studio Code, and PyCharm. One issue with notebooks is that they sometimes don’t play well with repositories because they save too much information, including metadata that doesn’t matter to anyone. That creates a problem when there are merge conflicts, as notebooks are cell-oriented and source code repositories such as Git are line-oriented.
