Serialization and Deserialization

Serialization represents values in a format that can be stored or exchanged. Deserialization reads that representation and reconstructs values. The goal is to preserve the logical record, not arbitrary memory addresses or the runtime’s internal object layout.

A customer record might contain an ID, a name, and an amount. A JSON serializer writes field names and values as text; UTF-8 then represents that text as bytes. These are different layers: serialization describes the record, while character encoding describes its text. Binary formats can represent records without making the whole payload human-readable.

Reading valid JSON does not establish what a field means. An amount of 1200 could mean dollars or cents. A missing field can mean something different from an explicit null. A schema and the accompanying contract specify types, required fields, units, and acceptable values. Consumers must also agree on changes: adding a field, renaming one, or changing its type may affect different readers differently.

A round-trip check serializes a value and reads it back, comparing the properties the contract promises to preserve. For an identifier, that might include leading zeros and every digit. An integer sent to a reader that converts all numbers to binary64 may lose digits, even though both sides parse the JSON successfully. Use a text representation when readers cannot preserve the required numeric range, and test an actual consumer rather than only one program reading its own output.

Reference: RFC 8259: JSON. For worked storage examples, see How Computers Store Data.