Whose Breakfast is it, Anyway?
Posted by Rick DeNatale Thu, 28 Dec 2006 13:12:00 GMT
I just ran across this article, via O’Reilly Ruby.
Ohloh is a web site which analyses open source projects by monitoring commits. The article is making a case that PHP is somehow better than Ruby (and Rails), is “eating Rails for breakfast,” because:
- About 18% of all new code added to open source projects in 2006, or at least those monitoried by Ohloh) is written in PHP, compared to something like 2% in Ruby. The trend is upward for both PHP and Ruby although PHP’s slope is steeper
- About 10% of open source developers contributed code in PHP, vs. Ruby contributors who accounted for about 2.5%. The trend here is downward for PHP. It peaked at around 14% in 2004. Python and Perl are slightly below PHP and are in decline as well. Ruby rose gradually in 2004 and has been level in this metric since then.
- 5% of open-source projects started in 2006 are being written in PHP. This is about the same as Python. Perl struggles here at about 1%. All three languages saw a decline in project starts since 2002, although PHP and Python have increased gradually over the past two years.
This last set of statistics is one hint that Ohtoh’s analysis might be slightly flawed. In contrast to PHP (and the others), 16% of new open source projects in 2006 are being written in Ruby, and the number has been growing since the start of their charts in 2001.
Which begs the question of whether these numbers really show that fewer developers are getting more done in Ruby with fewer lines of code than in PHP. If this is indeed the case, I’d say that it’s really PHP, not Ruby and Rails which is in the cereal bowl.
Lines of code has always been a lousy measure of code, and often it correlates negatively with quality. Ruby programmers pride themselves on keeping their code DRY, avoiding unnecessary duplication of code. I’ve always gotten great satisfaction working on refactoring phases of projects which resulted in a net reduction in the total number of lines of code.
In any event, such arguments are far from convincing. Despite the question of how the data has been gathered, processed, and interpreted, there’s the larger question of the validity of arguments based on populariey.
