I’ve been doing a fair amount of work in Amazon’s OpsWorks, in many ways an elegant service. Once you have a set of chef recipes provisioning properly, you’ll want to create an AMI for the layer in question so that you don’t have to wait through a long setup process again. Unfortunately, doing this in OpsWorks can be frustrating since the instructions for making it happen are scattered across four entirely different documents. For my own sanity I made a checklist of all the steps necessary to create an AMI on an instance store EC2 server: this is that checklist for anyone else who might find it useful.
Interviewing Symonds & Son
I’ve had a rash of posts recently about Symonds & Son, and I was intending to write a piece more pertinent to Rails or programming rather than the business side of things… but recently I was approached to do an interview about working with my family. In answering the interviewer’s questions, I thought the results were interesting enough to post on my blog – hopefully I’ll also be able to post the interview itself when it’s completed!
Supercharge Your Rails Development Environment
I love trying to improve my development environment. Efficiency is important to me: I spend hours and hours a day programming, and even a 1% or 2% productivity boost would provide a massive time savings over the course of a year. Or, as xkcd so pithily put it:
Of course, I’ve probably obliterated any potential savings by spending so much time experimenting, but now that I’ve done it you don’t have to! Listed here is everything important to me to get my development environment zipping along. Hopefully you’ll be able to find something useful here too that makes your programming experiences a little bit faster.
Programming the New Family Business
In my previous post, I told the story of how I accidentally fell into consulting and turned it into a successful business. But what does a successful consulting business – specifically, my successful consulting business – actually look like?
Symonds & Son isn’t named to sound cute and hipstery. I’m the son; I’m in business with my father, and I recently hired my mother to be my assistant and bookkeeper. The company’s name is intended to evoke images of traditional, reliable family businesses, where trust, excellence, and a reputation for quality products are their cornerstones. These companies, traditionally centered around carpentry and other hand-made professions, are my models for how to run Symonds & Son. We are, if you’ll forgive me using the phrase, a bespoke development shop, where our great output and superb relationship with our clients are our highest priorities.
But building my profession into a company – and a company with my parents, no less – hasn’t been an easy road. There have been a lot of compromises we’ve had to make to get this ship sailing straight. Here, I’ll speak to how I made this new parent/child relationship work… and how I hope to continue making it work for years to come.
Rails Consulting for Fun and Profit
2013 was a great year for me specifically, and my web development shop (Symonds & Son) in general. Though I initially fell into consulting accidentally, I’ve aggressively parlayed it into a successful business – and my only regret has not been doing so sooner! A lot of developers I know are on the fence about striking out on their own. I’m going to lay out how 2013 changed me from a full-time employee to owner of my own business, and in doing so hopefully persuade a few people that the benefits to being in business for yourself far outweigh the risks.
Cryptolalia, for Creating Cicada 3301 Ciphertext Puzzles
After reading yet another article on Cicada 3301 on Hacker News a few weeks ago, I was struck by inspiration. If I wanted a whole gaggle of people on the Internet to compete for – well, for some unknown goal – apparently all I needed was ciphertexts of arbitrary complexity and arcane places to hide them. Arcane places I had, but what I didn’t was a good way to generate all the sorts of codes that Cicada 3301 employed.
So I decided to make it.
Don’t Code Defensively
As an engineer, we’re responsible for creating things that people care about deeply. Whether you’re programming something as important as a car’s operating system or as silly as an iPhone game, our code touches lives. It’s no wonder, then, that the people who interact with our products feel strongly about them – and possibly about us, as well.
This can sometimes be a good thing. It’s always touching to receive a heartfelt letter of thanks and admiration from a customer. But much more commonly, you’ll hear complaints, demands, and anger.
Under a deluge like that, it’s easy to become disillusioned and bitter. I see it all the time: programmers who hide in their shells at the slightest provocation. They make excuses, shift blame, and refuse to own their actions. I think this attitude is ultimately cowardly – if you’re willing to take credit for the good things you’ve done, you have to “take credit” for the bad stuff too. And cowardice of this sort contributes to making bad code, since your goal is no longer to to solve your organization’s problems, but to shirk responsibility.
That’s what I call coding defensively – an entirely separate subject from defensive programming. Coding defensively mires you in your codebase. It’s hard to go forward because you don’t want more responsibility, and you can’t go back because it would be admitting error. And it’s a self-reinforcing cycle: once you start becoming defensive, it’s difficult to stop.
I was a defensive coder too. I struggled to own the results of my actions: I would refuse to discuss solutions in a sensible, level-headed way. It took me a long time to change my approach, and as part of doing so I wrote a list of maxims summarizing how I wanted to think and act. When I feel myself slipping I go over them to refresh my resolve. I’m publishing them here in the hope that they’ll help someone else as much as they’ve helped me.
What Makes a Good Programmer Good?
I’ve worked with a lot of programmers over the years – some of them super amazing, and some distinctly lackluster. As I’ve had the pleasure of working with some very skilled individuals recently, I spent some time thinking about what I admire in them. What makes a good programmer so good, and a bad programmer so bad? Or, to mangle English a little bit, what makes a good programmer good?
Based on my experiences, being a great programmer has nothing to do with age, education, or how much money you make. It’s all in how you act and, more deeply, how you think. I’ve noticed a consistent set of habits in the programmers I admire. More than knowledge of their chosen language, deep understanding of data structures and algorithms, or even more than years of on-the-job experience – the way they communicate, the way they conduct themselves, and the way they approach programming speak volumes as to their amazing level of skill.
Certainly there’s more to being a good programmer than anyone can enumerate, and I wouldn’t judge any programmer solely based on the presence (or lack) of these practices. But I do know wisdom when I see it, and when I see a programmer expressing these traits, I think, “This person really knows what they’re doing.”
Sidekiq + Houston: Production Ready
Three months ago, I wrote Sidekiq + Houston: Persistent Apple Connection Pooling. The code I included there initially worked great but over time all the APN connections I had established would break and not restart themselves appropriately. To correct this issue, I wrapped the APN connection itself in a class that was more resistant to failure. To help those who are using Sidekiq and Houston together in production, here’s the code I used to do so.
Helper Methods in Sprinkle
Recently I’ve been using sprinkle a lot in a large client project. Sprinkle is server provisioning software, akin to Chef except much lighter. It’s most directly akin to rubber, but rubber’s biggest advantage is its pre-built recipes: it’s a little finicky to sensibly extend, and those only work well on EC2. Sprinkle is built for extension, customizability, and platform agnosticism, but comes with no recipes at all by default. Tradeoffs!
Sprinkle (and rubber) are very different from most other server provisioning software I’ve used – on the one hand, by leveraging Capistrano for server communication (or SSH or Vlad if you prefer), it remains extremely light and focused on just provisioning. But on the other, it inherits most of Capistrano’s downsides too: primary among them is that it’s easy to repeat yourself if you’re not careful. So I wanted to post a quick tip for other people using sprinkle on how to DRY it up just a little bit.