What MessagePack Is

MessagePack ("msgpack", invented 2008) is a binary format for the same data model as JSON: strings, numbers, booleans, null, arrays, maps.

Wire format goal: compact + fast. No schema needed.

JSON:      {"name":"Alice","age":30,"active":true}
           38 bytes

MsgPack:   83 a4 6e 61 6d 65 a5 41 6c 69 63 65 a3 61 67 65 1e a6 61 63 74 69 76 65 c3
           25 bytes (~33% smaller)

vs JSON:

• Smaller: integers as binary (1-9 bytes vs decimal text); shorter strings.
• Faster: no string parsing for numbers; byte-oriented.
• Type-preserving: int vs float distinct, binary data native.
• No schema: like JSON, every message self-describing (vs Protobuf needing schema).

vs Protobuf:

• No schema: simpler, but no field renaming / evolution rules.
• Slightly less compact (Protobuf optimizes for known schemas).
• Faster setup: just install library; no code-gen.

Real systems:

• Redis MessagePack: optional binary format.
• Pinterest, Treasure Data, Fluentd: heavy users.
• WebSockets, RPC frameworks: where JSON is too verbose.
• MongoDB BSON is similar in spirit.

vs other schemaless binary:

• CBOR (RFC 8949): IETF-standardized similar format.
• BSON: MongoDB-flavored.
• Smile: Jackson's binary JSON.
• Ion (Amazon): binary + text formats.

We'll build a parser/encoder. Simple but covers all common types.

Reference: https://github.com/msgpack/msgpack/blob/master/spec.md