My current pet project is a Boilerplate OpenGL-ES 2.0 Android project in java. I don't want to go all the way and build an engine, I'm far from being experienced enough, but at least it's a great way to sort the good and the bad in all the things I've done already.
Loading of textures, creation and loading of materials, loading of shaders, linking of programs, creation and buffering of VBOs... there's a ton of things to be done before you can even see anything on your screen. I'd love to speed this up for the first launch, and then let the code be overhauled according to each and everyone's specific needs - OpenGL is too low-level to be wrapped in a full-fledged library (that would be an engine...), and I'm a bit fed up with copy-pasting my own bugs from previous projects, so a boilerplate with very few features (and errors!) will do.
Ah yes. I wrote a game (text-based "sports" sim) in Python and wxPython roughly eight years ago that still has a small, but hardcore community of players. I started on a "version 2" of the game about six years ago. Anytime I want to learn a new language I try and implement "version 2" using the new language. As a result I've learned Ruby, JavaScipt, ActionScript, Scala, Clojure, Lua, Haxe, and a ton of frameworks/libraries associated with the previous languages. It's been an absolute blast, but I still haven't released "version 2" yet.
My own pet project is a browsergame: 2 years in I still don't have anything playable because I start over all the time.
It began with Ruby, Sinatra and MongoDB; the last version before I deleted it was with Rails, Haml and ActiveRecord.
Next time I'll go with Rails, Slim and Sequel.
I don't think I'll ever release anything, but it's fun to work on it and learn new tech!
I agree. I think I take the reverse approach -- I maintain a list of project ideas that I would actually use were I to build it -- and when I get the free time, I can see what new languages / libraries / techniques I can apply to implement one.
Loading of textures, creation and loading of materials, loading of shaders, linking of programs, creation and buffering of VBOs... there's a ton of things to be done before you can even see anything on your screen. I'd love to speed this up for the first launch, and then let the code be overhauled according to each and everyone's specific needs - OpenGL is too low-level to be wrapped in a full-fledged library (that would be an engine...), and I'm a bit fed up with copy-pasting my own bugs from previous projects, so a boilerplate with very few features (and errors!) will do.