developer.* Blogs - Apgar, Metrics, and CVS - Comments

Google Video

Steve Benz — Fri, 17 Nov 2006 19:05:57 -0800

That was quite an interesting talk! At one point the speaker calls out data that seems to conclusively show that, almost all of the time, the committer turns out to be a useless predictor of whether a commit will be buggy or not.

Honestly, I'm not entirely surprised. I think good programmers can be as error prone as bad ones. What I think distinguishes a good programmer from a bad one is the survivability of their code. That is, the good coder will check in reusable components that last in the product, unchanged, for many years. The bad programmer will treat each new challenge as a separate entity and never write much code that is reused. When their code does persist, it's usually because it becomes dead code.

Bugs via SCM

Siddharta Govindaraj — Tue, 14 Nov 2006 19:48:13 -0800

Hi Pat, there was a video on Google video about predicting bugs using source control information. Not sure if its the same one. The video is here.

I detect a self-contradiction

Edward G Nilges — Tue, 14 Nov 2006 17:40:11 -0800

See my extended post "Squaring the Circle"

similar ideas

Pat Eyler — Tue, 14 Nov 2006 16:34:57 -0800

In his OSCon 2006 presentation 10 Tools for Open Source Projects, Karl Fogel presented some similar ideas about using tools to evaluate bug reports and repository commits to screen potential commit-bit cendidates or other folks who would be good to groom for a bigger role in an open source project. I can't find any links to his talk right now, but I think it was released as an mp3 somewhere.

Apgar, Metrics, and CVS

Steve Benz — Tue, 14 Nov 2006 13:47:33 -0800

When I read the article Daniel referred to in his Apgar score post, and I came away with a different lesson. My central observation was that simply by tracking the Apgar score, doctors and hospitals became more interested in reviving babies that they might previously have diagnosed as stillborn.