20 May 2012
I’ve been thinking recently about how hard it feels to contribute to an open source project and by that I don’t mean that open source makes it hard for you, au contraire, some of them try to make it as easy for you to do it as possible, sites like github, codeplex, google project hosting, etc., give you and developers the tools to build an open source project and try to foster a community around it.
I think one of the main motivations people have to contribute for creating and contributing to open source projects is having a need. You need something and you simply do not find something that solves your need in the market (or you can’t pay for it) and create an open source project, or maybe there’s a project somewhere that almost solves it and it just needs a little push so you contribute. You need to use it to be able to be an effective contributor, and most of the time, you use it because you need it.
You can find a lot of examples of this but the canonical example is Linus Torvalds starting Linux because he wanted to run Unix on his 386 and his only option at the time was MINIX and he was frustrated by its licensing.
Nowadays of course, most people don’t have a need to build an OS from scratch (do they), so they build other fancy things like Redis or Ruby On Rails which again, were started to scratch a need and people have been adopting in en masse because they had that pain too.
Funny thing is, we look at these examples and think that a successful contribution or a successful project has to be a complex piece of software that solves the needs of millions, or contributing to a project involves being a genius, nothing far from the truth, lots of open source projects thrive with the contributions of every day developers like you and me caring enough to research a bit and be interested in helping.
When I was in college, a few years ago, I used FreeBSD on my personal computer which was a bit of a problem because the school machines and my professors used Windows machines. So when I was developing something (mostly in C back then) it had to compile and be able to run in Windows. For the most part this isn’t very hard to do if you stick to ANSI C which suffices for most of your average school projects. That is, until you try to do something fancy…
As part of my AI class, we analyzed several combinatorial optimization algorithms and tried to solve different problems with them, one of the problems of choice was the 8 queen puzzle which basically consists on coming up with arrangement of 8 queens in a regular chess board so that no two queens can attack each other.
At the end of the semester I decided to present as a final project a version of all the algorithms we saw trying to solve this problem but instead of having only print outs of numbers or strings as a solution, to actually have a 3D chess board and the queens moving from place to place as the algorithms worked on the problem so it would be more visual and more interesting to look at. I coded this using C and OpenGL under FreeBSD and since I needed to compile it under Windows too I used GLUT to handle my windowing and input needs.
The way I went was using the Mesa 3D library which at the time was at its 5.* version, but in FreeBSD’s ports the available version was 3.* (called Mesa3) which for some reason would not work in my machine, the window opened and the program ran and I got output on my terminal but the display would be always black and I struggled with this for a while but it really pissed me off that with a hand compiled current (at the time!) Mesa 5.* version it would work flawlessly. The reason it pissed me off is because I have never liked manually installing things in my *nix systems, if you want to remove them you have to look for the files, read the Makefile, etc. I always try to use the package manager or I feel my system as “unclean”.
So, in order to be able to use Mesa5 and still feel somewhat clean was to make it a package, so I did, I copied the Mesa3 port directory to a new Mesa5 directory and changed the Makefile (and I do not remember if I had to change the patch files a little bit too) to work with Mesa 5.*. I got my project running, able to compile correctly under FreeBSD and Windows and got an A+ on my AI class.
I also sent the files to FreeBSD (I do not recall what medium I used), I thought they were going to ignore them and I thought they did because I never actually got an answer.
Here’s the kicker, years later, as I was googling around my old email address for reasons I do not remember, I found out a series of emails about my commit!, they actually received it and considering making me the maintainer of a newly created mesagl port (mine was called Mesa5 I guess mesagl was created by another guy with my same problem) until another guy came with another update that was better. I never found out when it was actually happening because I wasn’t in the mailing list (which is one of the first things they ask you to do if you are going to commit stuff, my mistake).
So that’s the way I almost became a FreeBSD port maintainer. Maybe if I go back to FreeBSD at some point (using archlinux now) Ill try to get to maintain a port.