Open MichaelRawson opened 3 months ago
@ibnyusuf has to confirm, but I don't think NewCNF could at any time read HOL.
Right - I thought we USER_ERROR
ed with "NewCNF doesn't HOL yet" at some point in the relatively-recent past. But seems not, maybe we should put it back.
Ahah. We still do - try a "proper HOL" problem, and we bail out before CNF transformations with "Vampire is not HOLy". However PUZ091^5 is written in the THF dialect but it's actually a FOOL problem with no higher-order part at all, so Vampire continues and crashes another way somehow - does NewCNF support FOOL in principle?
On closer inspection it's not even FOOL - just propositional!
Perhaps somebody with NewCNF or HOL expertise knows what's going on.