Writing clean Bash code

The shell is a paradox. It is mysterious in its complexity and marvelous in its simplicity. You can learn the bare basics with just of couple hours of tinkering, but becoming a shell master takes years of consistent use, and even then it feels like you learn something new every day. On one hand, it has an ugly syntax, and trying to work with strings and arrays is beyond frustrating. On the other hand, there’s nothing quite like the zen achieved when you stitch together a half dozen shell commands to extract and format your data of interest, creating a pipeline that outperforms any Python or C++ code you could ever write.

Whatever your feelings about the shell in general, or about (the most popular shell) Bash in particular, you cannot deny that it is a rustic language with very poor debugging facilities that makes it very easy to shoot yourself in the foot. If you can avoid writing code in Bash, you should. In fact, as blogger Jeff Atwood says, if you can avoid writing code at all, you should. But if you’re a modern biologist then chances are that sooner or later the shell will be the best tool for a particular job. What do you do then?

Over the last few months I’ve really tried to look for ways to improve my shell code. Here are some of the resources I’ve found.

  • Defensive Bash Programming: The author of this blog post provides some suggestions for keeping Bash code clean and readable—often a real challenge when stitching together shell commands in a pipeline, for example. He also provides suggestions for organizing the code and using variables carefully to avoid common pitfalls. If I’m writing a simple Bash script with just a few lines, I’d probably ignore most of this advice, but the longer the script gets, the more important it is to take extra steps to be careful and make your intent clear.
  • Shellcheck: So I was telling the truth when I said there’s no debugger for Bash, but this tool does its best to remedy that. If you paste a shell script into Shellcheck, it will analyze the code and point out potential issues or pitfalls.
  • Unofficial Bash strict mode: This is probably my favorite. The Perl language has a strict mode that makes writing and debugging Perl code a whole lot less frustrating. Bash has no such strict mode, but there are some things you can do to enforce safer behavior in your bash scripts. This blog suggests placing two lines at the beginning of every Bash script, and then describes why this is a good idea in a variety of scenarios. I haven’t taken his suggestion fully to heart quite yet, but I have begun placing set -eo pipefail at the beginning of most of my Bash scripts, and I have already saved myself tons of heartache and wasted time by doing so.

So the next time you absolutely, positively have to write some shell code, consider referring to these resources to improve the experience!

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s