572751c56e
Before this commit, we were inferring whether an inductive/structure/class were meta or not. This was bad since the user had no clue whether the type was trusted (non meta) or not. Moreover, users could get confused by this behavior and assume the kernel was allowing trusted code to rely on untrusted one.
47 lines
2 KiB
Text
47 lines
2 KiB
Text
open list
|
|
|
|
meta definition e := environment.mk_std 0
|
|
|
|
definition hints := reducibility_hints.regular 10 tt
|
|
|
|
vm_eval environment.trust_lvl e
|
|
|
|
vm_eval (environment.add e (declaration.defn `foo []
|
|
(expr.sort (level.succ (level.zero)))
|
|
(expr.sort (level.succ (level.zero)))
|
|
hints tt) : exceptional environment)
|
|
|
|
meta definition e1 := (environment.add e (declaration.defn `foo []
|
|
(expr.sort (level.succ (level.zero)))
|
|
(expr.sort level.zero)
|
|
hints tt) : exceptional environment)
|
|
|
|
print "-----------"
|
|
open name
|
|
|
|
vm_eval do
|
|
e₁ ← environment.add e (declaration.defn `foo []
|
|
(expr.sort (level.succ (level.zero)))
|
|
(expr.sort level.zero)
|
|
hints tt),
|
|
e₂ ← environment.add_inductive e₁ `Two [] 0 (expr.sort (level.succ level.zero))
|
|
[(`Zero, expr.const `Two []),
|
|
(`One, expr.const `Two [])] tt,
|
|
d₁ ← environment.get e₂ `Zero,
|
|
d₂ ← environment.get e₂ `foo,
|
|
/- TODO(leo): use
|
|
|
|
return (declaration.type d)
|
|
|
|
We currently don't use 'return' because the type is too high-order.
|
|
|
|
return : ∀ {m : Type → Type} [monad m] {A : Type}, A → m A
|
|
It is the kind of example where we should fallback to first-order unification for
|
|
inferring the (m : Type → Type)
|
|
|
|
The new elaborator should be able to handle it.
|
|
-/
|
|
exceptional.success (declaration.type d₁, declaration.type d₂,
|
|
environment.is_recursor e₂ `Two.rec,
|
|
environment.constructors_of e₂ `Two,
|
|
environment.fold e₂ (to_format "") (λ d r, r ++ format.line ++ to_fmt (declaration.to_name d)))
|