Hi,

With PR #2068, I have added support for the binary representation for varlena types.

I would like to also add support for fixed-size types, but I am not sure whether the CBOR serialization makes any compression attempt that would turn an apparently fixed-size type into a de-facto varlena type.

I understand that several traits currently derived for varlena types would have to be implemented separately for these fixed-size types - other than that are there any other particular difficulties you foresee?

My plan is to add a decorator like #[internallength(usize)], derive all of the necessary types (including assert_eq(serialized.len(), expected_internal_length) every time I serialize or deserialize to catch errors early) and that should be about it.

Do let me know what do you think of it.

Best,
Luca