JSON
JSON (JavaScript Object Notation) is a lightweight, text-based format for interchanging structured data. It originated as a subset of JavaScript object literal syntax, but it is language-independent. Nearly every modern programming language can parse and produce JSON without needing a JavaScript runtime.
Two specifications define it. ECMA-404, published by ECMA International, fixes the
syntax. RFC 8259, from the IETF, covers the interchange format and registers the
application/json media type. The two are kept in alignment.
Data model
JSON is built on a small set of types, mirroring the primitives most languages already provide.
- Objects are unordered collections of name/value pairs, written in braces. Names are strings; values may be any JSON type.
- Arrays are ordered lists of values, written in brackets.
- Strings are double-quoted and Unicode.
- Numbers are a single numeric type. JSON makes no distinction between integers and floating-point values.
- Booleans (
true,false) and null round out the set.
The format is human-readable and self-describing, which is a large part of why it spread so quickly. It is also strict. There are no comments, no trailing commas, no bare identifiers, and strings must use double quotes.
Adoption
For much of the late 1990s and early 2000s, XML was the default interchange
format for web services. JSON overtook it as the dominant format for web APIs
because it is far less verbose and maps directly onto the native data structures
of most languages – objects, arrays, strings, and numbers – without a separate
parsing layer. Its close alignment with JavaScript was decisive on the web,
where the browser’s native JSON.parse and JSON.stringify made client-side
handling trivial.
Today JSON is the de facto payload format for REST and other HTTP APIs. It is also the base for several companion formats. JSON Schema describes and validates the structure of JSON documents. JSON-LD layers linked-data semantics on top of JSON, giving plain JSON objects a machine-readable meaning. JSON Web Tokens carry signed claims as JSON.
Trade-offs
JSON’s simplicity is also its limitation.
- Text, not binary. JSON is verbose relative to binary formats such as Protocol Buffers, and parsing it is slower. For high-throughput internal communication, binary formats and frameworks such as gRPC are often preferred.
- One number type. Because JSON inherits JavaScript’s single numeric type, integers above 2^53 lose precision when round-tripped through a JavaScript-based producer or consumer. Systems exchanging large 64-bit identifiers typically encode them as strings.
- No native schema. JSON describes values, not the shape they must take. Schema and validation are handled by the separate JSON Schema specification.
- No comments. The grammar forbids comments, which makes JSON awkward as a configuration format that humans edit by hand. Formats such as TOML and YAML fill that niche instead.