agda/agda#7674 eta-expand fields in record expression #2546
Conversation
Agda PR #7674 requires some implicit arguments explicitly in some record expression given since variance info got refined (invariant to non-variant) and hence solutions are no longer unique.
src/Algebra/Construct/Initial.agda
Outdated
|
|
||
| isMagma : IsMagma _≈_ _∙_ | ||
| isMagma = record { isEquivalence = isEquivalence ; ∙-cong = ∙-cong } | ||
| isMagma = record { isEquivalence = isEquivalence ; ∙-cong = λ {y = y} {v = v} → ∙-cong {y = y} {v = v} } |
There was a problem hiding this comment.
This checks with Agda master (2.8.0) but not with 2.7.0.1:
_x_48 ≈ y → _u_49 ≈ v → _x_48 ∙ _u_49 ≈ y ∙ v !=
{v = v₁ : Carrier} → x ≈ y → v ≈ v₁ → x ∙ v ≈ y ∙ v₁ because one is
an implicit function type and the other is an explicit function type
when checking that the expression
λ {y} {v} → ∙-cong {y = y} {v = v} has type Congruent₂ _≈_ _∙_
Apparently, in 2.8.0 we have fixed another bug but I could not find it now.
|
@andreasabel IIUC, as this is an |
Formally, yes. However, in this case the change is in its present form compatible both with current Agda and 2.8.0, and it does not change the value of any definition, only makes some arguments explicit, so I thought we could just merge this into |
agda/agda#7674 eta-expand fields in record expression
Agda PR #7674 requires some implicit arguments explicitly in some record expression given since variance info got refined (invariant to non-variant) and hence solutions are no longer unique.
Fixing Agda to commit 2ffb002 now in agda/agda#7674.
If this PR is rebased or squashed (rather than merged with merge commit) the commit hash there needs to be updated accordingly.