Build servers are for more than just building

Using a build server (such as CruiseControl doesn’t mean developers shouldn’t run local builds (even though broken builds aren’t really as serious as a lot of people make them out to be). So this raises the question: if developers run their build locally, what’s the build server for?
Continue reading “Build servers are for more than just building”

Fun with graphs – metrics for CruiseControl builds

Update: Woo-hoo… I got the second graph working the way I want. Check out the demo page linked below.

I’ve been playing with XPlanner recently, and I was impressed by the graphs that it could produce. So I knocked up something similar (using the same technology) for CruiseControl.

‘ve been playing with XPlanner recently, and I was impressed by the graphs that it could produce. So I knocked up something similar (using the same technology) for CruiseControl.

You can find a snapshot of it at http://www.users.on.net/robertdw/demo/demo.html… the data behind the two graphs is real, BTW (the reason there were so many broken builds at the start is because it was a brand new project, and the builds didn’t pass until there was something to compile), so this isn’t a fake in any way.

The two graphs I have got up are a pie chart showing proportion of good vs bad, and a time-chart showing number of good/bad builds each day. (What I want to do in the end is have the Y axis of the time chart be the time of day of the builds, so we see when builds of both types cluster, and how long they take to fix)

Obviously for the real thing, I would polish it up and place it in as another tab on the reporting web page. Once I’ve done playing (in a couple of days, maybe a week), I will be committing the result. However, I’d love feedback, and although I’ve sent a letter to the CruiseControl devel list, I thought I’d post this here as well to see what people think.

So… thoughts and feedback anyone? Suggestions? To quote Dr Crane: “I’m listening”. Drop a comment in or send me a letter.