Perfect practice makes perfect

I have recently discovered exercism.io, by Katrina Owen of Ruby Rogues fame.

Exercism is a great way to practice your code and get feedback on it from others. You work on one exercise at the time and re-submit for review as often as needed until your code is considered great code. You then move on to the next exercise.

It’s awesome because you get feedback on your code, but also because you get to give feedback on other’s code and get to benefit form seeing others code evolve.

So far, it’s been a really fun experience. Thank you Katrina.

I recently read a blog post recently about Conway’s Game Of Life: Since I had not written an implementation myself, I decided to give it a go.

It was a fun exercise. I did not use any external library other than ruby’s stdlib and used TDD for most of the classes. The interface uses curses. Not knowing how to test it, I took a page from Gary Bernhardt in Functional Core, Imperative Shell and didn’t test the shell at all :)

Here it is: My Take on Conway’s Game Of Life

I found myself writting a gem for internal use that needed to run in several different ruby versions and MRI. In addition, the projects that it would be used in had other dependencies which also meant that it would need to function with different versions of dependencies. Here is how I ensured I was testing all scenarios.

UPDATE: The benchmarks have been updated for new versions of ruby, jruby, puma and unicorn

Previously I wrote about the difference in performance between using thin, puma and unicorn on heroku’s cedar stack. At the time, heroku’s recommendations where to use thin as a server. They now recommend unicorn for better performance and concurrency.

Additionally, I received some very valid criticism on the previous benchmark: Puma is designed to be used on a ruby implementation that does not have a Global Interpreter Lock: Namely Rubinius or jRuby. With that in mind, I decided to test again.

I follow a certain number of ruby-related blogs and have noticed regular of calls to use Vim as an editor, usually with claims of unbridled productivity to follow. One one hand, I see the appeal: I use a bunch of UNIX tools and scripts in my daily work-flow and it makes sense to use a editor that works inside my terminal. I have tried a few tutorials, screen casts and what not, but I have never felt really comfortable. Something has always been off. I always assumed it was due to the big learning curve others talk about.

On the other hand, when I switched from TextMate to Sublime Text, there was almost no drop in productivity. Part of it has to do with similar functionality being mapped to the same keys, like command-t. However, I believe that there is more to it.