We address the mismatch between typing Felleisen's C operator of
type ¬¬A → A and the need for a non-⊥ toplevel. We compare the
approaches of Griffin, Krivine and Ariola-Herbelin in solving
this issue. We propose a more relevant typing for C than ¬¬A → A
(see also the conclusion of the
JFP07 paper).