Open knisht opened 3 months ago
Yeah, here you are defining two sizes that run through the same structure (SPN/SPM) but in opposite directions: The structure is supposed to be antitone in the first index and monotone in the second, but as you say, the structure is superpositioned between mu-nu and nu-mu.
Agda does not check whether the size assignment makes sense.
Consider the following data type:
Yet again an inconsistency with sized types, but I believe that this one is of a new kind. Both
spn
andspm
are legal, and∞ < ∞
is not abused anywhere.I think the problem here is that
SPM
andSPN
can be expressed neither asνY.μX.Y
nor asμX.νY.X
, which messes with the theory. It is unclear how to forbid such definitions though.