I used De Bruijn indices when attempting to prove correctness of a toy language of mine. It's an elegant solution, but I found them exceedingly difficult to reason about both intuitively and in proofs. That being said, I've yet to find a better system (that isn't just augmenting De Bruijn indices) for implementing any sort of lambda calculus.
No comments yet.