ah btw <@U4UGS5FC7>, there is one problem with the...

# arrow

[Eugenio]

ah btw @raulraja, there is one problem with the reified approach: even if we could know all the places where the function is invoked (and so all the types to provide), we can only generate one generic extension function that would have to disambiguate the types from an instance at runtime, where all the extra type information is lost...