Are best practices always the best?
I've been trying to be an active participant on the CS Educators StackExchange. You should try to as well!!!
I answered a question about using the command line which led to a back and forth. I gave an example of something I might do in class to illustrate the power of using the command line and linked to some of my past posts. There was some back and forth in the comments - some of the participants weren't happy with my code as they felt it was filled with poor coding practices.
I'm not going to comment on the discussion other than to say that I make no apologies for how I use the Linux tools and the command line as my methods have served me well over the years. I also make no apologies on how I teach my students as I seem to have by and large done right by them over the years as well.
The crux was that I was showing what I might type in a class without any of the class commentary. It wasn't meant to be a script in a file let alone reusable code or a full blown program.
I don't want to spend time here talking about "best practices" for shell scripting. There are plenty of resources a search engine away.
The question is that how universal are best practices and is it more important that we teach specific best practices or rather that we have a discussion with our students about best practices.
Some general best practices are probably pretty universal – "write clear code" but I've seen best practices that when followed religiously just don't always make sense:
- never break out of a loop
- one return statement per function
- always or never do anything in particular
- all code should be commented
- always use an ORM
to list a few.
So, best practices aren't always best. More like guidelines to think about when you're programming and then you and your team can make smart decisions.
So, when I wrote those lines over on StackExchange, was I really showing my students how to code badly? The code:
ls *gif | while read i do f=`basename $i .gif` convert $i -resize 200x200 smaller-$f.gif done
doesn't exist in a vacuum. We start with an ls to see the directory, we use the echo throughout to test things before we run it live and then we run it live in such a way that it can't be destructive. I've been using the command line this way for decades and while I've screwed plenty of things up, it's never been due to a bad command line shell practice.
The important part is the discussion with the class:
- would this be appropriate in a script file?
- what are the dangers
- why and when is it ok?
It's not our job to dictate to our students which best practices to follow but rather to prepare them to make smart decisions.
This year, I'm back to using C++ in class after maybe two decades. I'm not up on current C++ best practices. I was wondering what was best:
or even if we should use "\n" vs endl.
I did my research, presented the results to my students and we talked about what they should consider when making their decision.
Students don't need us to blindly spout best practices. They can read them online or in a book. They do need us to guide them towards critically thinking about them.