[Opinion] Kent Beck's function distribution experiment
I love this. Kent Beck asks a question, and it's a good question on the distribution of code lengths in functions. One hour later, Tudor Girba provides an answer with data directly from a 1.9 million LoC system. What isn't obvious is that at least 30 minutes of that hour was taken up in discussions with myself on a completely different topic - chapter 4 of Rewilding Software Engineering.
Which means he must have finished our work, wandered over and seen the question on LinkedIn and provided the answer in less than 30 minutes. Knowing Tudor Girba it probably took less than a few minutes. [Just found out from Tudor, that it took 3 minutes].
So, I'd like to run a second experiment. Can you look at answering the question posed by Kent Beck - provision of a distribution for length of functions - for a similar size codebase (e.g. 1.9M LoC). But when you do provide the answer on Kent's post, could you also record how long it took you to get to the answer (time to Answer) and how many Lines of Code there are in your system. My specific interest is in the time it takes you to answer Kent's question for large systems.
If it's going to take a very long time (i.e. more than 10 minutes) and you can't do this, could you simply guesstimate how long it would take (in minutes, hours, days or even weeks), how many LoC (roughly) in your system (or a note to say that would take too long to find out) and add those answers here rather than cluttering up Kent's post with comments. I'll consolidate both sets.
Originally published on LinkedIn.