Taco Casa OS — Android (Compose) project

This is a Kotlin/Jetpack Compose rewrite of taco_casa_os.py, structured as a buildable Android Studio project. The operational state is persisted through Jetpack DataStore and historical operational changes are captured in an event ledger.

How to build the APK

1. Open this folder (TacoCasaOS/) directly in Android Studio.
2. Android Studio can regenerate the missing Gradle wrapper files if needed; gradle/wrapper/gradle-wrapper.properties points at Gradle 8.7.
3. Let Gradle sync and resolve the Android, Kotlin, Compose, Serialization, and DataStore dependencies.
4. Build → Build Bundle(s) / APK(s) → Build APK(s).

Project layout

app/src/main/java/com/tacocasa/os/
  model/        — TacoCasaState and @Serializable supporting data classes
  data/         — versioned DataStore persistence + OperationalEvent ledger
  viewmodel/    — TacoCasaViewModel and operational mutations
  ui/theme/     — Compose theme
  ui/components/— reusable kitchen-ticket components
  ui/screens/   — Home, Prep, Inventory, Cleaning, Notes
  MainActivity.kt

app/src/androidTest/java/com/tacocasa/os/
  TacoCasaRepositoryPersistenceTest.kt — write/recreate/read regression tests

Persistence contract

Operational state is encoded as Kotlinx Serialization JSON inside a versioned PersistedStateEnvelope and stored in Jetpack DataStore. LocalDateTime values use an explicit ISO-8601 serializer. Unknown JSON fields are ignored so additive state-model changes remain backward compatible.

The state schema has an explicit version. Future releases should add migrations in migrateState() rather than changing the persisted contract blindly. Legacy raw TacoCasaState JSON is also accepted during the transition from the original placeholder implementation.

Event ledger

Every operational state mutation is observed by the ViewModel persistence loop. The repository persists an append-only bounded ledger of OperationalEvent records alongside the current state. Events cover shifts, rush changes, revenue, expenses, staff, inventory, kitchen orders, prep, cleaning, maintenance, notes, complaints, recovery, alerts, and metrics.

This creates the foundation for:

Observe → Understand → Predict → Recommend → Approve → Act → Verify → Learn

The current ledger is state-diff based. The next architectural step is to promote important events to first-class domain commands with explicit actor, approval, prediction, outcome, and verification fields.

Persistence test

TacoCasaRepositoryPersistenceTest.write_kill_recreate_read_compare writes a populated operational state, constructs a new repository instance to simulate process recreation, reloads the state, and compares the restored object with the original. A second test verifies the event ledger also survives recreation.

What's faithfully ported vs. adapted

• Core operational state and business actions remain represented in the Kotlin ViewModel.
• The CLI's numbered input() menu is replaced with Compose touch UI and navigation.
• State survives app restarts through DataStore instead of remaining only in memory.
• Historical state changes now survive restart through the event ledger, creating operational memory for later verification and learning.