Code archeology

I’ve been spending quite some time the past days digging around in old code (3+ years old with no or very little changes). It’s quite fun noticing what efforts pay off and which doesn’t when you go back to make chance to old stuff. I’m sure my observations aren’t generally applicable. We’re web developers and are keeping web platform (several pretty big websites on a common codebase) alive year after year.

Proper and sane variable- and function names pays off instantly. Incredibly short or misguiding function and variable names doesn’t. While it may be fun naming functions “doMagic” or likewise, chances that you – or someone else maintaining the code years after will probably not appreciate it at all.

Reasonable commenting on functions and tricky code parts often come to good use, but too much commenting is distracting. Some is good, and less is often more valuable.

Version control logs can be quite valuable, if they’re descriptive for the change made. Too often though, was the comment “misc. bug fixing”, “problem solved” or something which years later was of no use or value at all.

Documentation kept apart from the source code seem to have none or very little value. Not once have I bothered to look in the wiki-documents or any other documentation which was in the code. In the few instances I did, minor changes and bug fixes had changed the code to such an extend (while not updating the documentation) that it was misleading at best.

The only cases where the documentation apart seem to make a positive is database diagrams and descriptions. They do seem to have value.