Project 1 - Implement ActivityStreams extensions

[warriordog]

Project 1 - Implement ActivityStreams extensions

Overview & Background:

This project aims to implement full support for ActivityStreams extensions. Extensions are way to define custom properties and types that extend the base ActivityStreams and ActivityPub specifications. Each extension uses a unique JSON-LD context to define its changes and ensure that other applications can properly handle the custom data.

Extensions, while technically optional, are a key component of ActivityPub, especially for projects that aim to integrate with existing Fediverse software. They are also remarkably difficult to implement in statically typed languages like C#, due to their support for:

• Adding properties to a type at runtime. (dynamic typing)
• Objects of more than one type. (multiple inheritance)
• Properties, types, and even values that are remapped at runtime based on the JSON-LD context. (renaming / aliasing)

None of these features are supported natively in C#, which requires this implementation to emulate them. Doing this properly requires a careful balance between developer experience, spec compliance, performance, type safety, and complexity. A number of research stories have been created to ensure that all key concerns are properly evaluated before moving forward with a solution.

Goals

Goals of this project are (loosely in order):

• Extensions can be added by application code
• Extensions can be added by 3rd party libraries
• Support for any existing, real-world extension
• Ergonomic API (should be comfortable to use in the common case)
• Conventional API (should look like a normal C# API / type model)
• Simple to write custom extensions
• Support for any extension that extends one type
• Support for any extension that extends multiple types
• Support for any valid extension

[warriordog]

Current status:

• Model extended types (base AS Type + some extended properties)
• Model polymorphic types (object of two or more types simultaneously)
• Respect JSON-LD context while parsing
• Emit extension contexts while serializing
• Use pre-loaded contexts
• Follow inline contexts
• Download unknown contexts (stretch goal)