Tip of the session: in order to "unit test" your common Ant scripts, have a test project that you can run the scripts against. You can then assert that the required artifacts are created as expected.
There was a discussion by Joe of the Ant Script Library, an open source library of re-usable build targets
Ivan talked about the problem of putting too much programming logic into build scripts, and looked for input about the best way to avoid this problem. Jeff suggested that choreographing complex logic in Ant is probably best done inside a custom Ant task, instead of trying to set up targets, or using Ant Contrib.