Is there an idiomatic approach to smart constructors for structs, other than defining a function that wraps the original constructor?

How smart do you want it to be? struct supports a very limited auto value, so if that suffices, you can use it.

There’s https://docs.racket-lang.org/struct-plus-plus/ if you need it to be more advanced .

Generally speaking, I’m thinking about constructors that verify invariants for fields and, perhaps, do some preprocessing before building the struct value.

However, in the case of what I’m currently working on, I just realized that smart constructors are not the right thing.