A "declaration has metavariables" error

[mik-jozef]

Prerequisites

Please put an X between the brackets as you perform the following steps:

• Check that your issue is not already filed.
• Reduce the issue to a minimal, self-contained, reproducible test case.
• Test your test case against the latest nightly release.

Description

The following code

structure ValVar (D: Type) where
  d: D
  x: Nat

def Set (T : Type) := T → Prop

structure Salg where
  D: Type
  I: Nat

def natSalg: Salg := { D := Nat, I := 42 }

inductive HasMem (salg: Salg): Set (Set (ValVar salg.D))
| hasMem
  (set: Set (ValVar salg.D))
  (isElem: set ⟨d, x⟩)
:
  HasMem salg set

def valVarSet: Set (ValVar Nat) :=
  fun ⟨d, x⟩ => x = 0 ∧ d = 0 ∧ d ∉ []

def valVarSetHasMem: HasMem natSalg valVarSet :=
  HasMem.hasMem
    valVarSet
    (show valVarSet _ from ⟨rfl, rfl, nofun⟩)

throws (kernel) declaration has metavariables 'valVarSetHasMem'.

There is a list of things that make the issue go away (which is also why I couldn't come up with a more specific title -- I've no idea what the underlying cause is :sorry:):

• removing x from ValVar, along with its occurences in the rest of the program
• removing I from Salg
• removing the salg param from HasMem & replacing occurences with natSalg
• removing d ∉ [] from the definition of valVarSet
• replacing HasMem.hasMem valVarSet with HasMem.hasMem _ in valVarSetHasMem.
• replacing show valVarSet _ from with show _ from, show valVarSet ⟨0, 0⟩ from, or removing it entirely.

Expected behavior: The code throws no errors.

Versions

Version: Lean 4.14.0-nightly-2024-11-01
Platform: live.lean-lang.org

Impact

Add 👍 to issues you consider important. If others are impacted by this issue, please ask them to add 👍 to it.

[Komyyy]

We have the similar issue:

open Function

theorem my_mul_right_injective {M₀ : Type} [Mul M₀] [Zero M₀] {a : M₀} (ha : a ≠ 0) :
  Injective fun (x : M₀) ↦ a * x := sorry

-- (kernel) declaration has metavariables 'mul2'
def mul2 : { f : Nat → Nat // Injective f } := ⟨fun x : Nat ↦ 2 * x, mul_right_injective nofun⟩

[kmill]

Note: this issue's example doesn't typecheck on a recent Lean, I think because of the changes in #6024. In the tests for #6414 I added the following unification hint to make up for the difference:

unif_hint (s : Salg) where
  s =?= natSalg
  |-
  Salg.D s =?= Nat

Unfortunately, this covers up the issue in 4.14. I figure it's still worth including the example in #6414 as a regression test, but it would be better if we could observe that this issue has been fixed.

testing on stable mathlib live-lean, which is currently 4.14