Yes, I've done this before, but it's not appropriate here. We're doing distributed processing on hundreds of pairs of files coming from one source. In the process, a certain amount of wheel reinvention has occurred, I'm sure.