The best serialization formats, according to hackers

9 serialization formats recommended in Hacker News comments over the past year, ranked by how many separate people recommend each one.

All serialization formats

Protocol Buffers (protobuf)Schema-based binary format with compatible changes, compact messages and generated code in many languages. Critics cite language design flaws and that messages need the schema to read.77recommend
SerdeRust's default library for converting data to and from formats like JSON, praised for pleasant parsing. Critics mainly cite slow compile times from heavy code generation.22recommend
Cap'n ProtoFast binary serialization and RPC system that people pick as a simpler, zero-copy alternative to Protobuf. Caveat: output can be large and the C++ code is less ergonomic.20recommend
ProtobufSchema-based binary format with good tooling and forward and backward compatibility across many languages. Critics most often flag its silent default values.19recommend
FlatBuffersFast binary serialization format that reads data without parsing, often recommended over Protocol Buffers; critics say its output can be large and bloated.16recommend
Lite³Fast zero-copy access to indexed, JSON-like data that can be edited in place, though critics say its wire format is not JSON-compatible.14recommend
RX random-access JSON formatDevelopers recommend it as a compact, random-access format for embedded data, with fast key lookups. Critics say it is not a drop-in JSON.parse replacement and the docs need work.11recommend
AvroRecommended as a compact, schema-based alternative to XML and JSON for RPC and stored data. Critics note data needs a copy step before use and enforcement needs extra tooling.10recommend
Apache ForyFree open-source serialization framework recommended for object graphs and high throughput. Critics note it relies on internal JDK APIs and question its benchmark data.6recommend