Introducing JSON Structure - A Better Schema

JSON Schema is a validation language, not a data definition language. This talk introduces a new specification designed to finally fix that.

#1about 5 minutes

Why JSON Schema fails as a data definition language

JSON Schema's complexity with validation constructs like `allOf` and `anyOf` makes it a poor choice for defining data types, leading to a stalled standard.

#2about 2 minutes

Introducing JSON Structure as a type definition language

JSON Structure refactors JSON Schema into a strict type definition language with required types, named objects, and new built-in primitive types for better code generation.

#3about 2 minutes

Using new compound types like map and tuple

JSON Structure introduces formal compound types like `map` for dictionaries and `tuple` for creating efficient, compact wire formats for time-series data.

#4about 4 minutes

Defining discriminated unions and organizing types

Use the `choices` keyword for formal discriminated unions and organize schemas with namespaces and a restricted `$ref` for cleaner mapping to programming languages.

#5about 3 minutes

Managing dependencies with imports and explicit inheritance

Separate file inclusion from type referencing with `$import` and use `abstract` and `extends` for clear, unambiguous type inheritance.

#6about 4 minutes

Extending schemas with companion specifications

Add functionality through optional companion specs for internationalization, scientific units, currencies, and even re-introducing validation rules from JSON Schema.

#7about 2 minutes