GitLab

We are switich away from De Bruijn indices because
1) it's hard to read terms with De Bruijn indices
2) they are making lots of things slower

And improve the proof of the fundamental property

A new formulation allows for doing more work inside HOSL.

Γ ⊨ e ≤log≤ e' : τ <=>
∀ Δ vvs ρ,
  spec_ctx ρ -∗
  □ ⟦ Γ ⟧* Δ vvs -∗
  ⟦ τ ⟧ₑ Δ (e.[env_subst (vvs.*1)], e'.[env_subst (vvs.*2)])

sans examples

We still need to prove the many admitted lemmas in rules_binary module.

The way the binary relation is defined, the binary fundamental lemma simply
doesn't hold. I overlooked the fact that we need to enable CAS only for types
where equality is checkable.

commit 392b7b43
Author: Amin Timany <amintimany@gmail.com>
Date:   Tue May 3 21:25:10 2016 +0200

    Finish using new iris with "proof mode"

    In Fμ and Fμ_ref we do support reduction under Fold.
    In fact `Unfold (Fold v)` is reduced to `v` if and only if v is a variable.

commit 9825e341
Author: Amin Timany <amintimany@gmail.com>
Date:   Mon May 2 20:35:57 2016 +0200

    Prove fundamental lemma of stlc is proven

    Change the Fμ to make the operational semantics reduce under Fold.
    Fundamental lemma for Fμ is partially proven (up to App).

In Fμ and Fμ_ref we do support reduction under Fold.
In fact `Unfold (Fold v)` is reduced to `v` if and only if v is a variable.

- ownI_proper for inv
- later_contractive for later_ne
anymore. Instead we use (contractive_proper _) and (contractive_ne _).

Instead we can use (contractive)