Going through the book I bought after listening to the Python Bytes podcast. Perfect as it was topic I had to deal with in my current project at work.

Also a good occasion for me to practice correct formating, docstrings etc as I transition from working on small-medium scale data analytics projects running in Jupyter Notebooks to a more public setting where documentation and readability must prevail.

Reading status: ongoing

Chapter 1 : Getting started with pytest

So far, nothing new.

Need to get better at virtual environments.

Chapter 2 : Writing Test Functions

First point: explanation of how to package a script is what I didn’t know I was looking for. Part of what is not taught in Data Science curriculum (see notes on Appendix 4).

Basic information about test functions. Testing if exceptions are raised (is it possible to test if exceptions are not raised? Seem to not catch if expression is caught in a try except clause.)

Need more practice with parametrize. What is more pythonic: using pytest.mark.parametrize or declaring variables in the test function? Learning along the way so my code is a mess.

Chapter 3 : pytest fixtures

To be read

Appendix 1 : Virtual environments

Hear about them too many times but never took the time to actually learn them. Never had to work with a package manager either (npm, composer). Chapter goes throught the bare minimum but at least it covers what I need to know (and the difference between Win and Un*x)

Appendix 2 : Pip

To Be Read

Appendix 3 : Plugin sampler Pack

To Be Read

Appendix 4 : Packaging and Distributing Python Project

How much I needed that ! Different infrastructures, what is need and where and so on. Perfect.

Following reading the chapter, I was looking for a way to automate the structure and settings. On Twitter, Rodrigo Fuentealba mentionned PyScaffold and that was perfect. I have been playing with it locally and have just started a small development at work based on it. It covers all I wanted by I still need to dive into the settings and the various arguments to really understand what is possible. I currently understand 30-40% of what it does and that far too little for me to be satisfied.