You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Equational reasoning works as advertised: equational-reasoning x = y by p computes to p, but it relies on refl being a strict left unit with respect to path composition, since the syntax expands to refl ∙ p.
This doesn't work anymore for e.g. equivalence reasoning, where equivalence-reasoning X ≃ Y by e computes to e ∘e id-equiv, not e as claimed.
The way this reasoning syntax currently works is that there's an X-reasoning_ operator which binds strongly and creates the left unit, and then the steps apply transitivity. I believe it can be reimplemented so that the steps bind strongly, and then X-reasoning_ would be the identity, but that's just speculation at this point.
The text was updated successfully, but these errors were encountered:
Equational reasoning works as advertised:
equational-reasoning x = y by p
computes top
, but it relies onrefl
being a strict left unit with respect to path composition, since the syntax expands torefl ∙ p
.This doesn't work anymore for e.g. equivalence reasoning, where
equivalence-reasoning X ≃ Y by e
computes toe ∘e id-equiv
, note
as claimed.The way this reasoning syntax currently works is that there's an
X-reasoning_
operator which binds strongly and creates the left unit, and then the steps apply transitivity. I believe it can be reimplemented so that the steps bind strongly, and thenX-reasoning_
would be the identity, but that's just speculation at this point.The text was updated successfully, but these errors were encountered: