Zulip Chat Archive

Stream: lean4

Topic: declaration has metavariables


view this post on Zulip pcpthm (Feb 01 2021 at 10:47):

I read somewhere that Lean4 improved the elaboration algorithm and we can use the field notation in more places so:

#check (fun n => n.succ) Nat.zero

works in Lean4.
But the expression above doesn't work when used inside an inductive type definition:

-- > (kernel) declaration has metavariables 'Test.mk'
inductive Test : Nat  Type
| mk : Test ((fun n => n.succ) Nat.zero)

Is it a bug?

view this post on Zulip Sebastian Ullrich (Feb 01 2021 at 11:11):

Fixed in https://github.com/leanprover/lean4/commit/d4dc54a724

view this post on Zulip Simon Winwood (Feb 01 2021 at 23:30):

That commit broke this (I think):

inductive SortedMap {α : Type u} {β : Type v} [LessOrder α] : List (α × β)  Prop
  | nil : SortedMap []
  | cons :  (k : α) (v : β) (l : List (α × β)),
           SortedMap l 
           l.Forall (λ (kv : (α × β)) => k < kv.fst) 
           -- N.B., HasLessOrder combined with this Forall implies
           -- no duplicate keys.
           SortedMap ((k,v)::l)
deps/galois_stdlib/src/Galois/Data/List.lean:54:10: error: typeclass instance problem contains metavariables
  LessOrder ?m.2558

view this post on Zulip Simon Winwood (Feb 01 2021 at 23:31):

This works in before this patch


Last updated: May 07 2021 at 12:15 UTC