tributary is a system for creating collections of end-to-end encrypted data.
A tributary collection has two key components:
- the app: code that organizes, maintains, etc the database and presents it in a user friendly format
- the stream: encrypted database (pglite) replication log that can be replayed to recreate the database
For example, for notes we are going to create a "scribe" app that maintains documents. Other apps may be more appropriate for e.g. photo or location / presence collections.
This project is organized into several components:
- supabase - database migration and serverless functions for the remote tributary server
- tributary-cli - a typescript cli command useful for debugging and testing
- tributary-client - a typescript library that wraps PGLite with persistence guarantees
- tributary-catalog - a react application for listing and managing all user collections.
- apps - where we maintain official apps
- scribe - a markdown document editor, indexer, and linker
- We write the very minimal backend on top of supabase.
- The core frontend database is pglite.
- For everything frontend, we default to typescript and vite. We use vitest for testing.
- For crypto, we default to tweetnacl / tweetnacl-ts. Good wrappers are not
available for these in rust, so we need to use the
ed25519_dalekandsha2crates directly.
- ALL COMPONENTS MUST BE THOROUGHLY TESTED
- WE NEVER USE MOCKS FOR TESTING. We prefer fakes where necessary that can be easily substituted for the associated clients for thorough integration testing.
