As this day draws to a close for me, I find myself wondering how much actual effort this little project has been so far. I’m a sucker for useless statistics. I hope you are, too… otherwise the following might bore you.

It’s fairly hard to quantify the effort I invested into this project in terms of development time. I can see that the first subversion commit was on the 11th of July, just over five weeks ago. But I haven’t been working on this continuously, and I certainly didn’t stick to regular working hours.

If I had to come up with an estimate, I’d say I spent about two thirds of those days working, not counting weekends. So say 5.5 hours per day, for about 5 * 5 working days… that’d be 137.5 man-hours, or roughly 3.5 man-weeks at fourty hours each. Seems about right.

But time isn’t the only metric. Let’s see what ohcount has to say about my lovingly hand-crafted code and markup.

                          Ohloh Line Count Summary                          

Language          Files       Code    Comment  Comment %      Blank      Total
----------------  -----  ---------  ---------  ---------  ---------  ---------
java                 35       5125       1521      22.9%       1681       8327
xml                  24       1364        447      24.7%        206       2017
----------------  -----  ---------  ---------  ---------  ---------  ---------
Total                59       6489       1968      23.3%       1887      10344

Nice. That’s about 6.5k lines of code (LOC), and about 2k lines of comments. That means I’ve coded at not quite 50 lines of code per hour, and 15 lines of comments per hour. Or 2k LOC/man-week.

Ok, ok, I won’t go on. We don’t want this to get ridiculous.

But I find metrics like that interesting insofar, as the COCOMO software cost estimation model assumes a program with about 10k LOC is produced, on average, at a rate of 3.2k LOC/year1.

Whoever came up with that model clearly operates in a world very different from my own, or that of most other engineers I’ve worked with2.

  1. Disclaimer: I’ve got that number from this blog post, which may very well be very, very wrong. []
  2. And yes, I know that one has a tendency to produce code faster on a fresh codebase. This is just an extreme example of something I’ve wondered about when working on far more mature codebases as well. []
Comments Off

Comments are closed.