serializer

A serializer is a software component that converts in-memory data structures or objects into a format suitable for storage or transmission. A corresponding deserializer reconstructs the original objects from the serialized representation. Serialization enables persisting application state, communicating between processes or services, and caching data.

Serializers may output binary or human-readable text. Text-based formats include JSON, XML, and YAML, while binary formats include Protocol Buffers, Thrift, Avro, and MessagePack. Some formats are self-describing, carrying data types and structure, while others rely on an external schema or code to interpret fields. Binary formats often emphasize compactness and speed, with schemas guiding compatibility across versions.

Use cases include web APIs, file storage, inter-process communication, and message queues. Serialization may preserve object graphs and references or serialize only data values for reconstruction as new objects. Handling of cycles, aliases, and identity requires careful library features or design choices within the format.

Key considerations include compatibility and schema evolution, performance, data size, and support for streaming large objects. Some systems prioritize forward and backward compatibility, while others favor speed and compactness. Streaming serializers process data incrementally rather than buffering entire objects.

Security concerns arise when deserializing untrusted input, as some formats or libraries can execute arbitrary code during reconstruction. Best practices include using well-maintained, safe serializers, validating schemas or types, and avoiding execution of arbitrary code during deserialization.

Examples by language include Java’s built-in Serializable (often discouraged); Python’s pickle (powerful but risky); and JSON serialization with libraries in many languages; modern binary formats such as Protocol Buffers or Avro are often preferred for performance and schema enforcement.