How joreilly/confetti works
Project type and purpose
• Confetti is a full-stack, Kotlin Multiplatform (KMP) GraphQL application. It consists of:
• A GraphQL backend (Ktor/Apollo) with a data-import microservice and a datastore layer.
• Multiple Compose Multiplatform clients: Android, iOS (SwiftUI host), Desktop, Web (WASM+JS), Wear OS, Automotive OS/Android Auto, and Android Car OS.
• Shared KMP libraries (business logic, Apollo client, models, DI) and protobuf definitions.
• CI/CD scripts (GitHub Actions), build-logic plugins, and Terraform for infra.
File-structure analysis
• Top-level clients:
• androidApp, iosApp, compose-desktop, compose-web, wearApp, automotiveApp, common/car
• Shared code:
• shared (KMP business logic + Apollo), proto (shared proto definitions)
• Backend services:
• backend/service-graphql (GraphQL server), backend/service-import (data importer), backend/datastore (datastore abstraction), backend/terraform (infra as code)
• Build & deployment:
• build-logic (custom Gradle tasks), .github/workflows (CI/CD pipelines), fastlane (iOS/Android store publishing)
Key README insights
• Uses Apollo GraphQL for client/server schema sharing.
• Clients talk to GraphQL endpoint at confetti-app.dev/graphql.
• Data import service populates the datastore (Firebase Datastore or Cloud Datastore).
• CI/CD deploys backend to Cloud Run/App Engine; clients to app stores.
Main components
• Android App
• iOS App (SwiftUI host)
• Desktop App (Compose Desktop)
• Web App (Kotlin WASM + JS)
• Wear OS App
• Automotive OS & Android Auto App
• Android Car App
b. Shared library (Kotlin Multiplatform):
• Business logic, models, Apollo client, DI
c. GraphQL Backend service:
• HTTP API (Apollo/Ktor)
• Resolvers → DataSource layers
d. Data Import microservice:
• Pulls from external conference data sources (Sessionize, DeepAI, etc.)
• Writes to the datastore
e. Datastore layer:
• backend/datastore module abstracts Google Cloud Datastore (or Firestore)
f. Infrastructure & CI/CD:
• Terraform provisioning (Cloud Run, App Engine, Firestore)
• GitHub Actions pipelines (build/test/deploy)
• Fastlane for mobile store publishing
g. External services:
• Gemini API (optional recommendations)
• Firebase Auth & Analytics
• Google Play Services (sign-in)
Relationships & interactions
• GraphQL Backend → Datastore module → Cloud Datastore/Firestore
• Data Import service → Datastore module → Cloud Datastore/Firestore
• Terraform → provisions Cloud Run/App Engine, Datastore
• CI/CD pipelines → build & deploy code to clients and backend
• Clients & backend share schemas via Apollo GraphQL & proto definitions
Architectural patterns & principles
• Modular KMP layers: UI layers per platform, shared business logic.
• Microservices: GraphQL API service vs import service vs datastore abstraction.
• Dependency injection via Koin (in shared and platform modules).
• Infrastructure as code (Terraform) and automated pipelines.
• GraphQL schema serving as contract between frontends and backend.