Although it was shown that the "full contamination" was not really full during the list discussion, the series improves things without incurring extra parsing cost, and here is a test to check that. Signed-off-by: Junio C Hamano <junkio@cox.net>