RE: Proposal for new last call issue: Some unprocessed headers should stay

I agree that such a table would be a great addition to the spec.

From a purely editorial POW, I wonder whether a non-normative appendix
would be appropriate given that the exact rules are already elsewhere in
the spec. That way, we avoid the confusion of which version is "the one
that rules them all" :)

Henrik 

>Since this seem to be found of interest, here's a further update 
>with "mustUnderstand" added.