Mar 12, 2018
Recently I've been teaching more coding than ever, both to coworkers and to students in my bootcamp. While they're all catching on and getting results, its made me step back and realize, modern day programming is not straightforward and with all the various parts, often is just plain confusing.
Python is a popular programming language so it feels simple enough to tell a new programmer, "Hey open up a text editor and write the following commands"
At this point the you feel accomplished, the student feels accomplished, but then it quickly wears off as both of you realize this program doesn't do anything useful. So you start trying to explain the next steps, but where do you start?
There's data types in Python, such as lists, dictionaries, etc. There's also the concept of iteration and control loops. But DRY is a thing so you quickly have to explain functions, but then you get into scope, and with Python there's whitespace formatting to cover as well.
Somewhere in there you fit in a lesson about shell commands, and depending on whether they're in Windows or a POSIX client you wave your hands about OS file systems.
All the code shouldn't be in one file though or it's confusing, so then you explain Python import machinery, but modules have their own nuances. Somewhere in there you ramble on about functional design vs OOP.
Well now the student has something more complex working and they're thrilled but they want to deploy it. So now you need to teach them about git, but its also important that they know about isolated environments through virtualenv, or conda. A conversation ensues about dependencies and why pinning versions for imported packages might be necessary.
After all this they finally get their code packaged up and put into a repo. Their code works and is available but man it's ugly so you start explaining the importance of code quality, and how PEP8 is a thing, and why spaces after single line comments and camel casing are oh so important. A linting package is installed a bunch of messages appear on the console after you show they how to run it.
They didn't know about unit testing though so you go through a lecture on Test Driven Development. Their monolithic logic needs to be refactored.
And now their code needs to be deployed, so you ask them hard questions of how it will be used. If it's going to be a local package did they make sure it worked across all operating systems and python versions?
They instead want a webapp, which eliminates most of the issues above, sweet! You ask them which web framework they want to use but now they need to learn about WSGIs and API design. An explanation about networking and HTTP response codes fits in there as well.
Then you need to talk about Continuous Integration and explain various deployment strategies from Virtual Machines to Docker Containers to self hosting versus cloud hosting.
And if they're doing machine learning be sure to cover at least a semester's worth of statistics.
Going from starter code to production is a lot of steps and for people that are new to programming it can be very frustrating because the code seemed to work 10 steps ago. If you're a seasoned programmer take a second to step back and realize how many moving parts there are to deploying a "maintainable production ready" app/library, and how to someone new it can be very difficult to see the forest for the trees! There's a lot of steps and concepts between Hello World and a robust library or application. When teaching someone new just take a breath and remember what it took you to gain that understanding as you guide your junior programmer. They will appreciate the help.