Code Review Coincidences
May 2nd, 2008
I discovered during my Wednesday trip to the MathWorks that they do extensive code reviews in house. Coincidentally, Guido van Rossum sent mail to the Python 3000 list the next day saying that the code review tool he built for Google isn’t going to be released, but one designed to work with Subversion is, and Christian Hammond offered free Review Board hosting for Summer of Code projects. The study that Smart Bear Software did at Cisco convinced me that code reviews can actually work (I was sceptical for a long time, as I though they fell into the “nice, but people won’t actually do it” bucket). Maybe their time has finally come?
More studies should be done like this one. There’s plenty of papers on team productivity in manufacturing, civil engineering, etc. in economic literature, backed by statistical analysis. Alas, the results from those studies may not necessarily be applicable to software development, since building bridges is unlike building software.
I actually wanted to write a paper for an econometrics course on whether agile software development has any effect on project success rates, but found that there’s very little empirical research in this topic, and data is almost non-existent. Basis would’ve been this article davidfrico.com/rico07k.pdf, but I’m not sure about credibility of the researcher, and his data is not publicly available.
As the author of the Smart Bear study, I completely agree that more studies should be done.
Also, although of course we sell the most popular code review tool (http://codecollab.com), I’m really happy to see all these other projects like Review Board and Reitveld appear on the scene, especially since the open source community is already so open to code review. In fact, almost all the large open source projects are fully reviewed, usually in the form of patch-submission.
Anything that helps people do more code review without wasting time is a good thing!