> Is your analysis in any way similar to how PMD defines it's cyclic > complexity metric? I always thought that PMD did a good job of > warning developers of potentially 'smelly' code.
No, it has nothing to do with McCabe's cyclomatic complexity, which basically tries to quantify the complexity of the control flow in a method (i.e. how many different paths there are through if-statemetns, loops etc). Basically what I'm doing is lookin at compilation dependencies among source files. So if A.java depends on B.java, and B.java depends on C.java, and C.java depends on A.java, then A.java and B.java and C.java are all in a dependency cycle. Graphically: - -> A -> B -> C -- ` | \_ _ _ _ _ _ _ _ _/ So in Ant-1.6.5 there are 117 .java files involved in a big dependency cycle. That is the height of the biggest red bar. The yellow bars show the distribution of another metric I have devised CRSS which aims to quantify how "flat" the structure of a system is. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]