minicbor

• IpAddr,
• Ipv4Addr,
• Ipv6Addr,
• SocketAddr,
• SocketAddrV4,
• SocketAddrV6.

Depends on minicbor-derive-0.18.0 (see below).

⚠️ Breaking ⚠️ Several generic implementations of Encode and Decode did not override the default is_nil and nil methods. As a result, if a type T implements Encode and Decode with implementations of is_nil and nil, wrapping the type in a Box or Cell or encoding it via reference would revert to the default implementation. This has been fixed and the following impls now override the defaults:

• &T
• &mut T
• Box<T>
• Cow<'_, T>
• Cell<T>
• RefCell<T>

With this fix, these types have a nil value if T has one whereas prior versions would not.

The Encode implementation of RefCell no longer uses try_borrow, but borrow.

⚠️ Breaking ⚠️: The Encode and Decode traits are now parameterised by a context type and the context value is passed as another argument to Encode::encode and Decode::decode (see merge request [21][mr21] and issue [26][i26] for details). Implementations of these traits that do not make use of the context need to be generic in the type variable and accept the context parameter, e.g. instead of

struct T;

impl Encode for T {
    fn encode<W: Write>(&self, e: &mut Encoder<W>) -> Result<(), Error<W::Error>> { ... }
}

impl<'b> Decode<'b> for T {
    fn decode(d: &mut Decoder<'b>) -> Result<Self, Error> { ... }
}

one would now write:

struct T;

impl<C> Encode<C> for T {
    fn encode<W: Write>(&self, e: &mut Encoder<W>, ctx: &mut C) -> Result<(), Error<W::Error>> { ... }
}

impl<'b, C> Decode<'b, C> for T {
    fn decode(d: &mut Decoder<'b>, ctx: &mut C) -> Result<Self, Error> { ... }
}

⚠️ Breaking ⚠️: The type data::Cbor has been removed. To write pre-existing CBOR bytes use Encoder::writer_mut with Write::write_all and to access raw CBOR bytes from the decoder use Decoder::input.

⚠️ Breaking ⚠️: The AsRef impl for Encoder has been removed. Use Encoder::writer instead.

⚠️ Breaking ⚠️: The legacy module and feature flag have been removed.

Decoder::input has been added to get a reference to the input bytes.

The newtypes minicbor::encode::{ArrayIter, MapIter} have been added to encode any clonable iterator as a CBOR array or map.

Several new methods have been added to Decoder and Encoder to work with contexts:

• Decoder::decode_with
• Decoder::array_iter_with
• Decoder::map_iter_with
• Encoder::encode_with

These correspond to the existing variants without the _with suffix which do not accept a context and fix the context type to (). Note that generic implementations of Decode and Encoder must therefore use the versions which accept a context parameter.

Other additions include the crate-level functions:

• encode_with
• decode_with
• to_vec_with

⚠️ Breaking ⚠️: The encoding of IP addresses changed (see commit fac39d5a). This affects the following types:

• std::net::IpAddr
• std::net::Ipv4Addr
• std::net::Ipv6Addr
• std::net::SocketAddr
• std::net::SocketAddrV4
• std::net::SocketAddrV6

A new module minicbor::legacy is introduced which contains newtype wrappers for these types which continue to use the array-based encoding. Users can opt out of the new compact format by enabling the cargo feature "legacy" and importing the types from the legacy module.

A new type minicbor::data::Int has been introduced (see merge request [20][mr20]) to allow encoding and decoding of the whole CBOR integer range [-2⁶⁴, 2⁶⁴ - 1].

⚠️ Breaking ⚠️: As a consequence of adding the new Int type, a new constructor minicbor::data::Type::Int has been added to denote those (signed) integers that do not fit into an i64. Similarly the new constructor minicbor::decode::Token::Int captures those values.

• core::num::Wrapping
• core::sync::atomic::{AtomicBool, AtomicI8, AtomicI16, AtomicI32, AtomicI64, AtomicIsize}
• core::sync::atomic::{AtomicU8, AtomicU16, AtomicU32, AtomicU64, AtomicUsize}
• core::cell::{Cell, RefCell}
• core::ops::{Bound, Range, RangeFrom, RangeInclusive, RangeTo, RangeToInclusive}
• std::path::{Path, PathBuf}
• std::time::SystemTime