3f0acbbb48
This PR fixes how we determine whether a function parameter is an
instance.
Previously, we relied on binder annotations (e.g., `[Ring A]` vs `{_ :
Ring A}`)
to make this determination. This is unreliable because users
legitimately use
`{..}` binders for class types when the instance is already available
from
context. For example:
```lean
structure OrdSet (α : Type) [Hashable α] [BEq α] where
...
def OrdSet.insert {_ : Hashable α} {_ : BEq α} (s : OrdSet α) (a : α) : OrdSet α :=
...
```
Here, `Hashable` and `BEq` are classes, but the `{..}` binder is
intentional, the
instances come from `OrdSet`'s parameters, so type class resolution is
unnecessary.
The fix checks the parameter's *type* using `isClass?` rather than its
syntax, and
caches this information in `FunInfo`. This affects several subsystems:
- **Discrimination trees**: instance parameters should not be indexed
even if marked with `{..}`
- **Congruence lemma generation**: instances require special treatment
- **`grind` canonicalizer**: must ensure canonical instances
**Potential regressions**: automation may now behave differently in
cases where it
previously misidentified instance parameters. For example, a rewrite
rule in `simp` that was
not firing due to incorrect indexing may now fire.
---------
Co-authored-by: Kim Morrison <kim@tqft.net>
Co-authored-by: Claude <noreply@anthropic.com>
63 lines
2.2 KiB
Text
63 lines
2.2 KiB
Text
/-
|
||
Regression test for https://github.com/leanprover/lean4/pull/12172
|
||
|
||
The pattern:
|
||
1. A class `MeasurableSpace` is used as both a class and explicit argument (via @)
|
||
2. `Measure.trim` takes a Prop proof `hm : m ≤ m0` and returns `@Measure α m`
|
||
3. A typeclass `SigmaFinite` depends on `μ.trim hm`
|
||
4. A function `myFun` has `hm` explicit and `[SigmaFinite (μ.trim hm)]` as instance
|
||
5. A section variable makes `hm` implicit
|
||
6. A lemma `myFun_eq` takes an explicit proof argument before the final arg
|
||
|
||
When calling `simp only [myFun_eq μ hs]`:
|
||
- Before #12172: `hm` is inferred, instance is found, works
|
||
- After #12172: Instance synthesis happens before `hm` is inferred, fails with
|
||
"failed to synthesize instance SigmaFinite (μ.trim ?m.XX)"
|
||
|
||
Workaround: `simp only [myFun_eq (hm := hm) μ hs]`
|
||
|
||
This pattern is used in Mathlib's MeasureTheory.Function.ConditionalExpectation.CondexpL1
|
||
-/
|
||
|
||
set_option autoImplicit false
|
||
set_option linter.unusedVariables false
|
||
|
||
universe u
|
||
|
||
class MeasurableSpace (α : Type u) where
|
||
dummy : Unit := ()
|
||
|
||
instance {α : Type u} : LE (MeasurableSpace α) where
|
||
le _ _ := True
|
||
|
||
structure Measure (α : Type u) [MeasurableSpace α] where
|
||
val : Nat
|
||
|
||
def Measure.trim {α : Type u} {m m0 : MeasurableSpace α}
|
||
(μ : @Measure α m0) (_hm : m ≤ m0) : @Measure α m :=
|
||
@Measure.mk α m μ.val
|
||
|
||
class SigmaFinite {α : Type u} {m0 : MeasurableSpace α} (_μ : @Measure α m0) : Prop where
|
||
sigma_finite : True
|
||
|
||
def myFun {α : Type u} {m m0 : MeasurableSpace α} (hm : m ≤ m0) (μ : @Measure α m0)
|
||
[SigmaFinite (μ.trim hm)] (n : Nat) : Nat := n
|
||
|
||
section
|
||
variable {α : Type u} {m m0 : MeasurableSpace α}
|
||
variable (μ : @Measure α m0)
|
||
variable {hm : m ≤ m0} [SigmaFinite (μ.trim hm)] {s : Nat}
|
||
|
||
theorem myFun_eq (hs : s > 0) (n : Nat) : myFun hm μ n = n := rfl
|
||
|
||
-- This should work (worked before #12172)
|
||
theorem test_implicit_hm (hs : s > 0) (x y : Nat) :
|
||
myFun hm μ (x + y) = myFun hm μ x + myFun hm μ y := by
|
||
simp only [myFun_eq μ hs]
|
||
|
||
-- Workaround with explicit hm also works
|
||
theorem test_explicit_hm (hs : s > 0) (x y : Nat) :
|
||
myFun hm μ (x + y) = myFun hm μ x + myFun hm μ y := by
|
||
simp only [myFun_eq (hm := hm) μ hs]
|
||
|
||
end
|