Closed dselsam closed 3 years ago
Currently, applicative and monad cannot be aligned because the seq_left and seq_right classes changed to require the discarded computation to return punit. This is not a major problem, but there is some category theory in mathlib that uses applicative and monad and so the two barely-different versions would need to persist in mathlib4. I think backporting the change would be simple (here is a starting point: https://github.com/dselsam/lean3/commit/7e8ead2c0d371ddacef7db30e7dd68d52f4b768f) but Mario thinks it would be a change for the worse (https://leanprover.zulipchat.com/#narrow/stream/270676-lean4/topic/mathport/near/226098511) and wonders if the rationale for the change even still applies (https://leanprover.zulipchat.com/#narrow/stream/270676-lean4/topic/mathport/near/226098127).
applicative
monad
seq_left
seq_right
punit
There is an open PR by Sebastian reverting the lean4 change: https://github.com/leanprover/lean4/pull/311 This should resolve the issue.
leanprover/lean4#311 was merged and alignment went smoothly
Currently,
applicative
andmonad
cannot be aligned because theseq_left
andseq_right
classes changed to require the discarded computation to returnpunit
. This is not a major problem, but there is some category theory in mathlib that usesapplicative
andmonad
and so the two barely-different versions would need to persist in mathlib4. I think backporting the change would be simple (here is a starting point: https://github.com/dselsam/lean3/commit/7e8ead2c0d371ddacef7db30e7dd68d52f4b768f) but Mario thinks it would be a change for the worse (https://leanprover.zulipchat.com/#narrow/stream/270676-lean4/topic/mathport/near/226098511) and wonders if the rationale for the change even still applies (https://leanprover.zulipchat.com/#narrow/stream/270676-lean4/topic/mathport/near/226098127).