Applicative vs generative encoding — is the () barrier the only difference?

Status: waiting (needs collaborator confirmation)

The double-vision source encoding may need two mode-specific variants. The conjecture: the generative variant is identical to the applicative one modulo a () barrier that prevents path equality from leaking. Is that the only difference, or does generativity require more?

What would settle it

Either a proof that the ()-barrier variant is sufficient, or a counterexample where generativity needs additional structure. Flagged #waiting on collaborator input.

Double-Vision Recursive Modules · Source-to-Source Encoding Correctness

Sources

2026-06-04