lean4-htt/tests/lean/run/test_simp_reducible_class.lean
Leonardo de Moura c2ec2ecab1
fix: handle class projections in isNonTrivialRegular for backward.whnf.reducibleClassField (#12639)
This PR fixes the interaction between
`backward.whnf.reducibleClassField` and `isDefEqDelta`'s
argument-comparison heuristic.

When `backward.whnf.reducibleClassField` is enabled, `unfoldDefault`
reduces class field projections past the `.proj` form at `.instances`
transparency. This causes `isDefEqDelta` to lose the instance structure
that `isDefEqProj` needs to bump transparency for instance-implicit
parameters. The fix adds an `.abbrev` branch in `isNonTrivialRegular`
that classifies class field projections as nontrivial when the option is
enabled, so `tryHeuristic` applies the argument-comparison heuristic
(with the correct transparency bump) instead of unfolding.

Key insight: all projection functions receive `.abbrev` kernel hints
(not `.regular`), regardless of their reducibility status. Structure
projections default to `.reducible` status, while class projections
default to `.semireducible` status. The old code only handled the
`.regular` case and treated everything else (including `.abbrev`) as
trivial.

Also fixes two minor comment issues in `tryHeuristic`: "non-trivial
regular definition" → "non-trivial definition" (since `.abbrev`
definitions can now be nontrivial too), and "when `f` is not simple" →
"when `f` is simple" (logic inversion in the original comment).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-22 21:20:33 +00:00

56 lines
2 KiB
Text

module
/-!
# Test: `simp` with `@[reducible]` class field projections
When a class field like `X.x` is marked `@[reducible]`, `isDefEqDelta` unfolds it to `.proj` form.
`isDefEqProj` must then bump transparency to `.instances` when comparing the struct arguments,
since the projection's parameter is instance-implicit. Without this bump, `eq_self` fails because
instances like `instX a` vs `instX b` are stuck at `.reducible`.
-/
namespace SimpReducibleClassField
@[implicit_reducible] def a := 0
@[implicit_reducible] def b := 0
class X where
x : Nat
instance instX (n : Nat) : X where
x := n
-- Test 1: plain simp, semireducible X.x (works on master)
-- isDefEqArgs bumps to .instances for instance-implicit param of X.x
example : (instX a).x = (instX b).x := by simp
-- Test 2: plain simp, @[reducible] X.x
-- With backward.whnf.reducibleClassField = false: isDefEqDelta unfolds X.x to .proj form,
-- isDefEqProj bumps to .instances via withInstanceConfig.
-- With backward.whnf.reducibleClassField = true: tryHeuristic in isDefEqDelta applies the
-- argument-comparison heuristic, and isDefEqArgs bumps to .instances for instance-implicit params.
set_option allowUnsafeReducibility true in
attribute [reducible] X.x in
example : (instX a).x = (instX b).x := by simp
-- Test 2b: same as Test 2 with backward.whnf.reducibleClassField explicitly enabled
set_option allowUnsafeReducibility true in
attribute [reducible] X.x in
set_option backward.whnf.reducibleClassField true in
example : (instX a).x = (instX b).x := by simp
-- Test 3: simp [X.x] with semireducible args exposes stuck .proj node
-- reduceProjFn? unfolds X.x at .instances, but the .proj can't reduce further
-- because instX a' is not a constructor app at .reducible. This is expected:
-- the user explicitly requested the unfolding.
def a' := 0
def b' := 0
/--
error: unsolved goals
⊢ (instX a').1 = (instX b').1
-/
#guard_msgs in
example : (instX a').x = (instX b').x := by simp [X.x]
end SimpReducibleClassField