I occasionally will see test runs continue to run after I have Control-C killed the parent JVM. I believe this is a problem with how we are handling signal propagation to children. Could be an issue with the flags we pass to posix_spawn for the child's signals (e.g. it's not properly handling a signal from its parent to shut down).

Example I ran into today was running rake spec against reel and killing it shortly after it starts; the specs continue to run and belch out results into the interactive terminal.