2016-05-18 UTC
# 
willnorris part of the problem (I think) is that the parsing spec links to the value-class-pattern page for dt-* values. And I think that page is trying to cover both mf1 and mf2. that's mostly fine, but there are a few small changes I think