feat: simple type support

[beltram]

Hi,

I've crated this PR to add support for CBOR Simple types as defined in RFC 8949 Section 3.3.

I've roughly used the same approach you used for tags. I chose to fail in case a reserved simple type was used (0..=19) or (24..=31) and I think this is what causes the fuzzer to fail. He also seems not to differentiate one-byte and two-byte variants of simple types which causes some other errors in the fuzzer. I would like some pointers from your side to understand what happens under the hood and how to fix that.

Additionally, I wanted to take advantage of this PR to introduce a new variant Value::Undefined instead of Value::Simple(23). I think that's fine since the enum in #[non_exhaustive]. Let me know if that sounds reasonable to you.

Any help getting this PR fixed and merged will be very appreciated :)

Thank you

Closes #60

[cabo]

Nice! I was was alerted to this only now.
I don't think you should error out on the unassigned values.
We have reserved 24..31, so these are really not well-formed, but otherwise a decoder should present unassigned values to the application, because that may know more about a new assignment than a decoder library.

[beltram]

Alright, there was some missing pieces to make this work (especially when the SimpleType is in a Mapping or an Array for instance). This works now and the fuzzer is happy.

I agree with your point @cabo: however since this library was erroring on unassigned values it could subtedly break downstream consumers ; so this would require a new minor release.

@haraldh any thought on this ^ and on merging this ?

[haraldh]

@npmccallum your call

[cabo]

@npmccallum Any progress on this?

[rohanmahy]

@npmccallum @haraldh It sure would be nice if this functionality was incorporated. Then we would be able to use ciborium unmodified to implement SD-CWT.

[npmccallum]

@rohanmahy I'm not against this change. I do want to make you aware that I'm going to soon commit a new replacement for ciborium-ll (called ciborium-atom) that will fix a number of deficiencies. This will include better simple type support. But I'm not going to hold up your contribution until then.

You're currently serializing the simple type as an enum. I think a newtype would be a better option here.

https://docs.rs/serde/latest/serde/de/trait.Deserializer.html#tymethod.deserialize_newtype_struct
https://docs.rs/serde/latest/serde/ser/trait.Serializer.html#tymethod.serialize_newtype_struct

You can use the rename attribute to rename the struct to a magic name. But this should simplify the code significantly. I'll try to do a more thorough review this weekend.