Thinking about closed source as technical debt forced me to evaluate my whole computing life, taking a wholeistic look at all technical debt I have. I realized that closed source software is not the biggest problem facing me regarding technical debt. I realized there was a lurking debt that I hadn’t fully realized before: the state of my workstation.
Think about it. You probably have personal files on your computer that could get lost or stolen. If you read my blog, you probably have things heavily customized, set up the way you like them. What would happen if your computer died tomorrow? Each computer is a ticking timebomb, waiting to explode with headache.
Early last year (2012) bought I purchased a Macbook Air, which was to replace my Macbook Pro from 2007. I loved it and needed it, but it took me three weeks to set everything up the way I needed before it made sense to start using it full time.
What kind of debt accrues on the computers we use?
An unknown set of applications. What applications do you use? What would it take to get them back onto the computer? This is not much of a problem with large, obvious applications. Smaller programs, such as command line utilities installed with homebrew, are much easier to miss.
An unknown state of applications. What settings do I have? Which browser extensions? Where are all of the config files located for the unix utilities I use?
A lack of backups. Do I have backups? Are they sufficient? Are they secure? How could I know?
A lack of tests. I dont have a way to tell if all systems are working on my laptop. If I update a system gem, does that break anything? Without good tests, I can’t tell.
This is a problem I feel daily. As I have been trying to automate myself as much as possible, I hit these problems constantly. It is too easy to break a helper script that depends upon a file system structure, a version of RVM, or a specific shell command.
I also know I am not the only one with this problem. I once worked with very smart man who wrote many scripts to automate his work. However, whenever someone else on the team would try to use what he wrote, there was usually problem with them. These scripts expect certain applications to be present, certain libraries, and certain data paths. Moving the script to a different machine broke the silent expectations built into the scripts.
I have been using Chef to get my workstation under control. Pivotal has a few projects, such as Sprout, Sprout Wrap, and Soloist to help. If you also think this would be worth doing for yourself, you should take a look.