leanprover / leanprover/lean4

let elaboration more dependent than it should be

Open
#7,410 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

bug P-medium
Dominant language
Lean
Stars
9.2k
Forks
990
Avg merge
1d 17h
Merged PRs (30d)
175

Description

Consider this example

set_option pp.explicit true
/--
warning: declaration uses 'sorry'
---
info: P : Nat → Prop
m : Nat
⊢ @Eq Nat
    (let n := @HMul.hMul Nat Nat Nat (@instHMul Nat instMulNat) m (@OfNat.ofNat Nat 2 (instOfNatNat 2));
    @ite Nat (@Eq Nat n (@OfNat.ofNat Nat 0 (instOfNatNat 0)))
      (instDecidableEqNat (@HMul.hMul Nat Nat Nat (@instHMul Nat instMulNat) m (@OfNat.ofNat Nat 2 (instOfNatNat 2)))
        (@OfNat.ofNat Nat 0 (instOfNatNat 0)))
      (@id Nat (@OfNat.ofNat Nat 0 (instOfNatNat 0))) (@OfNat.ofNat Nat 0 (instOfNatNat 0)))
    (@OfNat.ofNat Nat 0 (instOfNatNat 0))
-/
#guard_msgs in
example (P : Nat → Prop) m :
  (let n := m * 2
   if n = 0 then id 0 else 0) = 0 := by
  trace_state
  fail_if_success simp (config := Lean.Meta.Simp.neutralConfig) -dsimp only [id]
  simp (config := Lean.Meta.Simp.neutralConfig) +dsimp only [id]
  sorry

  #version

The let looks very non-dependent – it binds a Nat, and the body of the let doesn’t care about the concrete value.

Still, as can be seen in the goal state above, while n is abstract in the condition of the ite, it has been unfolded in the Decidable instance.

This can have unwanted consequences: The let is considered a depenent let by simp, and (with -zeta), the body will not be simplified by simp, but only by dsimp.

Maybe uses of simp -zeta are rare; it did show up in the wf processing (issue #7408).

It would be desirable if elaboration (type class inference) would not unfold n if it doesn’t need to, if possible.

Versions

Lean 4.18.0-rc1

Additional Information

[Additional information, configuration or data that might be necessary to reproduce the issue]

Impact

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

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the reproducer in the issue and inspect elaboration, especially type class inference for the Decidable instance. Compare the behavior with simp -zeta and the wf processing mentioned in issue #7408. Done means elaboration avoids unfolding the let-bound n when its concrete value is unnecessary, while the reproducer shows the intended simplification behavior.

Written by the indexing model from the issue text.

Assessment

Domain
compilers
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.