Wednesday, November 05, 2014

...Or I Will Replace You With A Very Small Shell Script

Once again I remain convinced that there is no problem that is insurmountable when a possible first-draft solution includes a hacked-together bash script.

I spent the better part of an hour trying to bang my head against Windows, gave up, and grabbed my Mac laptop, opened a terminal, and accomplished the end-goal in 10 minutes, which included writing my own hacky first-draft do_the_thing.bash file. More and more, my solutions (including production-ready solutions in some cases) start life as do_the_thing.bash, and eventually evolve into a proper script with error conditions, failure modes, self-checks to prevent multiple runs, etc. But mostly, they start as do_the_thing.bash, which I blame a developer friend for doing in front of me and thereby teaching me terrible habits.

My own personal goal, after I get back from my convention trip, is to throw away do_the_thing.bash and rewrite the entire process, soup to nuts, in python (including checking for ruby and installing it if needed, among other steps)(and yes, I know, but ruby is also a thing I should be learning, and I feel like this is a good test case for that). I'll be leaning heavily on my more advanced dev resources (aka, my friends Rachel and Matt), but this is the first project I feel really excited about trying on my own.

When I was just making the transition from Tech Support person to DBA, I was often convinced that it was faster and easier for me to just execute a given command directly, rather than building a tool that executed a command. But as I've grown and matured, I've been put in positions where the execution of a particular command isn't the issue; it's the repeated execution of some set of commands that must be automated, and so I've built up a repository of script fragments stolen researched and designed as a library. It happened in bash and SQL; it's starting to happen in python, and I imagine as I get more dev experience it'll happen more and more in other languages. Often the hardest part of learning is making the transition from reinventing the wheel to recognizing that it's not always necessary to reinvent the wheel. Wheel-inventing is an important step; that's the point where I'm learning the "why" of things, not just the "how".



For now, though, I'm all about do_the_thing.bash. Or do_the_thing.py, or do_the_thing.js, or whatever. First build a skateboard. Then build a bicycle. Then build a motorcycle. (As a life-long motorcycle rider, we can just stop there.)