Page History

...

• Null type - indicated via optional fields
• Unions - you can get the equivalent with an object containing optional fields
• Repeated modifier - this is equivalent to lists
• Aliases - schemas copycat deals with are mostly generated automatically so aliases aren't really useful. They also introduce additional complexity in compatibility.
• Complex data types like dates/times, geographic coordinates, UUIDs, or other specialized data types that some systems might support. These fall into a couple of different categories:
  • Difficult to address comprehensively in a way that will be broadly compatible and satisfies different requirements (e.g. dates/times with all the complexity of timezones, leap days/seconds, different levels of granularity, etc).
  • Too niche to take on the burden of supporting them generally (e.g., geographic coords). Leaving the representation up to the connector is ok.
  • Only adds semantics to an existing type (e.g., UUIDs are just byte[16]).

Schemas

Schemas only have to exist as in-memory data structures, but to give a simple example we'll use a JSON-like representation here:

...