Lib.rs

messagepack-serde

MessagePack for no_std with serde.

Examples

use serde::{Deserialize, Serialize};

#[derive(Debug, Serialize, Deserialize, PartialEq)]
struct Data<'a> {
    compact: bool,
    schema: u8,
    less: &'a str,
}

let buf: &[u8] = &[
    0x83, 0xa7, 0x63, 0x6f, 0x6d, 0x70, 0x61, 0x63, 0x74, 0xc3, 0xa6, 0x73, 0x63, 0x68,
    0x65, 0x6d, 0x61, 0x00, 0xa4, 0x6c, 0x65, 0x73, 0x73, 0xa9, 0x74, 0x68, 0x61, 0x6e,
    0x20, 0x6a, 0x73, 0x6f, 0x6e,
];

let data = messagepack_serde::from_slice::<Data<'_>>(buf).unwrap();
let expected = Data {
    compact: true,
    schema: 0,
    less: "than json",
};
assert_eq!(data, expected);

let mut serialized = [0u8; 33];
let len = messagepack_serde::to_slice(&expected, &mut serialized).unwrap();
assert_eq!(&serialized[..len], buf);

Installation

Add this crate to Cargo.toml. If you want use this crate in no_std, disable default feature.

messagepack-serde = { version = "0.2", default-features = false }

Features

• no_std support
  If you want to use std::io::Read or std::io::Write, enable the std feature and use messagepack_serde::from_reader or messagepack_serde::to_writer.

• Flexible numeric serialization

  • Provides multiple serialization strategies:
    • Exact: Serializes numeric types exactly as provided.
    • LosslessMinimize: Minimizes size without losing information (default).
    • AggressiveMinimize: Aggressively minimizes values, including serializing floats with integral values as integers.
  • To deserialize arbitrary numeric values, use messagepack_serde::value::Number.

• ext format support

Design Decisions

Struct serialization format

This crate serializes Rust structs as MessagePack maps by default to preserve field names and allow flexible field ordering. Some other implementations (e.g., rmp-serde and MessagePack for C#) serialize structs as arrays by default.

To maximize interoperability, the deserializer accepts both map- and array-serialized structs. When an array is encountered, fields are read in the declaration order of the Rust struct.

// Example: deserializing a struct from an array and a map
use serde::Deserialize;

#[derive(Deserialize, Debug, PartialEq)]
struct S {
    compact: bool,
    schema: u8,
}

// [true, 0] serialized as a MessagePack array of length 2
let buf: &[u8] = &[0x92, 0xc3, 0x00];
let s = messagepack_serde::from_slice::<S>(buf).unwrap();
assert_eq!(s, S { compact: true, schema: 0 });

// {"compact": true, "schema": 0} serialized as a MessagePack map
let buf: &[u8] = &[
    0x82, 0xa7, 0x63, 0x6f, 0x6d, 0x70, 0x61, 0x63, 0x74, 0xc3, 0xa6, 0x73, 0x63, 0x68,
    0x65, 0x6d, 0x61, 0x00
];
let s = messagepack_serde::from_slice::<S>(buf).unwrap();
assert_eq!(s, S { compact: true, schema: 0 });

A major advantage of serializing structs as arrays is reduced output size. Smaller output sizes positively impact processing speed. Additionally, since you no longer need to search for properties within strings during deserialization, you can expect faster deserialization times as well. On the downside, this eliminates the self-describing nature of maps, making binary compatibility more fragile.

// Example: field was added and cannot be deserialized
use serde::Deserialize;

// A future version of the struct
#[derive(Deserialize, Debug, PartialEq)]
struct FutureS {
    compact: bool,
    // Strongly depends on field order; adding this may break old data
    awesome: Option<bool>,
    schema: u8,
}

// Older payload: [true, 0]
let buf: &[u8] = &[0x92, 0xc3, 0x00];
let s = messagepack_serde::from_slice::<FutureS>(buf);
assert!(s.is_err()); // cannot deserialize the array

// Older payload: {"compact": true, "schema": 0}
let buf: &[u8] = &[
    0x82, 0xa7, 0x63, 0x6f, 0x6d, 0x70, 0x61, 0x63, 0x74, 0xc3, 0xa6, 0x73, 0x63, 0x68,
    0x65, 0x6d, 0x61, 0x00
];
let s = messagepack_serde::from_slice::<FutureS>(buf).unwrap();
assert_eq!(s, FutureS { compact: true, awesome: None, schema: 0 });

This crate prioritizes robustness and binary compatibility, so structures are serialized in map format. For array-based serialization, consider representing the data as tuples or manually implementing Serialize implementations for each type.

// Example: serialize a struct as an array
use serde::Deserialize;

#[derive(Deserialize, Debug, PartialEq)]
struct S {
    compact: bool,
    schema: u8,
}

let val = S { compact: true, schema: 0 };
let mut buf = [0u8; 3];
let len = messagepack_serde::to_slice(&(&val.compact, &val.schema), &mut buf).unwrap();

// [true, 0] serialized as a MessagePack array of length 2
let expected: &[u8] = &[0x92, 0xc3, 0x00];
assert_eq!(buf, expected);

License

Licensed under either of

• Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
• MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)

at your option.

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.