c1ad6aa0db
This PR fixes a bug where `lia` was incorrectly solving goals involving ordered types like `Rat` that it shouldn't handle. The `lia` tactic is intended for linear integer arithmetic only. The fix adds `order := false` and `funCC := false` to `NoopConfig`, which is the base configuration for `CutsatConfig` (used by `lia`). Closes https://leanprover.zulipchat.com/#narrow/channel/113488-general/topic/releases.20of.20new.20Lean.20versions/near/564688881 🤖 Prepared with Claude Code --------- Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
38 lines
1.1 KiB
Text
38 lines
1.1 KiB
Text
/-!
|
|
# Test that `lia` does not use the order module
|
|
|
|
The `lia` tactic is for linear integer arithmetic only.
|
|
It should not solve rational number inequalities that require the order module.
|
|
-/
|
|
|
|
-- This should fail: lia should not handle rational inequalities
|
|
/--
|
|
error: `grind` failed
|
|
case grind
|
|
k : Rat
|
|
hk : 2 ≤ k
|
|
h : ¬1 ≤ k
|
|
⊢ False
|
|
[grind] Goal diagnostics
|
|
[facts] Asserted facts
|
|
[prop] 2 ≤ k
|
|
[prop] ¬1 ≤ k
|
|
[eqc] True propositions
|
|
[prop] 2 ≤ k
|
|
[eqc] False propositions
|
|
[prop] 1 ≤ k
|
|
[limits] Thresholds reached
|
|
[limit] maximum number of E-matching rounds has been reached, threshold: `(ematch := 0)`
|
|
-/
|
|
#guard_msgs in
|
|
example (k : Rat) (hk : 2 ≤ k) : 1 ≤ k := by lia
|
|
|
|
example (k : Rat) (hk : 2 ≤ k) : 1 ≤ k := by lia +order
|
|
|
|
example (k : Rat) (hk : 2 ≤ k) : 1 ≤ k := by grind_order
|
|
|
|
-- This should still work: natural number inequalities are handled by lia
|
|
example (k : Nat) (hk : 2 ≤ k) : 1 ≤ k := by lia
|
|
|
|
-- This should still work with explicit -order flag (no change in behavior)
|
|
example (k : Nat) (hk : 2 ≤ k) : 1 ≤ k := by lia -order
|