Tuesday, June 02, 2009

When overengineering fights getting things done

When I had some days off from work a few weeks ago I started to hack on a GNUmed plugin for cardiac device management. The idea was simple. Avoid all the mistakes done by the software I currently use and make it even better (TM). Back then I decided to document my steps toward the goal. Those of you having followed the series are aware that the planning stage went well. Having ideas and putting them on paper or mindmaps is the easier part. When the coding started it became quiet.

Let me explain why I think that is quite natural. One has to learn new tools. In my case that means learning python and wxglade. One has to learn about the existing code. In my case that is learning about database connections and getting patient information from the database. The single most critical issue howerver is the overengineering disease.

I work with a system that takes on pacemaker management. It has some flaws. Some of them I consider serious. Producing a new system involves finding and avoiding those flaws. I got trapped however by the desire to handle every use case in the first iteration of the plugin.

The only solution is to take a step back and reduce the initial feature set without cutting down on the possibilities. The plan is to get a working version out and enhance it gradually.

Another round of ICU shifts is coming up so this will have to wait for two weeks or so.

No comments: