27 Jul 2000 »
Opened my second SourceForge project today. I’m working on the secret project architecture. As soon as I have the High Level Architecture and Overview done, and as soon as the fantastic SF people approve my new project (pretty please!), I’ll post more details.
I’m going to undertake a social experiment with this project. The primary aim of the project is a bug free, high quality, CMM level 3 or 4 open-source project to replace an ancient but beloved tool on many Linux-like and Linux platforms. It will be architected to the nth degree before vi is opened or make invoked. By necessity, this requires a small or single person team to finish this part of the process before getting like-minded developers on to the job of actually cutting code. However, the process of developing the entire thing (besides the initial HLA) will be completely and utterly open in true bazaar style. I want to see if open source can produce a better tool when well known and well used software engineering practices are dolloped on top.
Software engineering best practice will be applied from woe to go to see if the defects in this tool, the MTBF and uptake are better than (say) fetchmail, which will be the reference project. I’ll write the experiment up for Linux Journal once 0.1 is released as source code to judge if it is a success. I’ll call the article “Post-bazaar software engineering” or something lame like that.