Code bloat

I found Jason Marshal's article on code bloat via Bob Congdon. It sparked some ideas of my own so I thought I'd write my own post.



Personally, I think this is one of the most interesting and challenging problems we face in the software industry.



Things like the widespread adoption of frameworks have helped reduce reinvention but paradoxically, across frameworks, a lot of code is duplicated. Personally I'm not convinced this sort of duplication is bad; even at more granular levels. Take for example two unrelated subsystems that deal with some of the same value objects. Is it better to couple these subsystems by sharing the value objects or is it better to duplicate them.



The problem today (programming in Java) is that unless there's an IS-A relationship between objects we need to build bridge objects that form the IS-A relationship. To better support this model of objects that are separate-but-the-same we really need a language that supports very late binding to individual methods (like Smalltalk or Ruby) or perhaps some sort of dynamic IS-A detection (I don't know a language that does this)



However, what I really want to see is some innovation around code and project visualization and organization. Tools like UML, Maven and Javadoc are cool in many respects, but I still find them all lacking. I look forward to better things to come.

Comments

Popular posts from this blog

Shark Crackers

Running roughshod or ripshod

Axis, Axes, Axii?