New JSON generator schema

Enter your suggestions in details:

Background

This issue is regarding the new format for the JSON generator.
It only pertains to the format of the JSON files, the implementation details will be discussed once a censensus is reached here.

Why a new format?

There are a handful of issues with the current format, with some of the main ones being:

• Maintainability
  • Without a pre-defined schema, it can be harder to tell where a property should be expected to go within the output
  • There isn't a great way to communicate changes to this schema when they happen
• Consumability
  • Users don't know what to expect without going through the generator's code or looking through all of the outputted JSON files
  • The current format represents some fields in unfortunate ways (i.e. Markdown being parsed in HTML for descriptions)

Relevant: DefinitelyTyped/DefinitelyTyped#70298, nodejs/api-docs-tooling#57

The new format

The newly proposed schema for json generator is available here.
An example of it being used for Buffer is available here.

The new proposed schema for the json-all generator is available here.
An example of it being used is available here.

Key Points

JSON Schema

The new formats have JSON schemas defined. This gives us three main advantages over the current format:

1. Consumers know what to expect
2. We can version the output files in a standardized way (via the $id property)
3. The schemas can be used to generate the types used within the JSON generator. This will help with maintaining the generator in the long run since we don't have to worry about them getting out of sync.

JSDoc Property Names

JSDoc keys (i.e. @name, @type) are used in the format.
This is mainly to make the files easier to consume.

TODOs

Here's what's left to be done with the new format:

• How do we want to represent sections that are just text (aka not defining any API)? Should these just be combined into their parent's description property? (Good examples for reference: the entirety of addons, Buffers and character encodings)