On an internal mailing list the question came up of how to take this stream of objects going in the pipeline and break it up into arrays of a given chunk size - so if you had 100 objects on the way in and decided to group in chunks of 10, you'd get 10