Tuesday, March 9, 2010

97 Things Every Programmer Should Know

It is probably a good time for me to plug the latest in the 97 Things series. It came out last month and I managed to get a story of my included:

97 Things Every Programmer Should Know, edited by Kevlin Henney.

This is another great work containing a wide range of writers interested in sharing their experiences and understandings. Something our industry desperately needs.

For this addition I threw in one of my favorite stories from around the age when I really learned programming. I was lucky enough to have a mentor, which made a huge difference to my abilities. I had no real idea what I was doing when I started.

You can work as hard as you want, but you can't get beyond just pounding out the obvious until you gain access to a larger knowledge base. That is, until you climb up on other's shoulders, you're just getting the same view as everyone else. You're just learning to do what they did already.

We may produce tonnes of reference material, and have pretty good online Q&A forms, but they are really only band-aides to help programmers fix things when they go wrong. Our biggest problems come from the way coders are building their systems. From the way they are just belting out code.

A mass of ugly poorly structured code may not be immediately visible, but you can always tell the poor construction from the disorderly interfaces, annoying over-complexity and quirky behavior. The true nature of code always shines through!

Things won't change until we find ways to teach the "higher" principles to new coders; there is far more to programming than just assembling massive lists of instructions. Things won't change until people understand that flailing away at their code is unnecessary, unproductive and just adds to the problem; brute force always produces messy systems. Things won't change until we learn how to pass on our skill sets; left on their own, most new programmers will just pound out the same messes over and over again.

That is why the 97 Things series is such a great thing for an industry that is clearly having trouble trying to grow. A huge number of our software development problems stem from the fact that each new generation of programmers poorly reinvents the same wheels, while strategically avoiding the real problems frozen into our mammoth backward-compatible tar pits.

In an age were we have access to massive data, powerful machines and can connect to computers from anywhere, it is the only our software that continues to be a let down.