# Plan Redesign: Waterfall → Milestone-Based Development

**Date**: 2026-03-03

## Problem

The original 9-phase sequential plan had three issues:

1. **Premature abstraction** — LedController trait designed before hardware validation
2. **Phases too large** — multi-session phases with unclear stopping points
3. **Hardware dependencies unclear** — couldn't tell what to work on with partial hardware

## Decision

Restructured into milestone-based groups organized by hardware availability:

| Group | Hardware Needed | Milestones |
|-------|----------------|------------|
| **A** | XIAO M0 + LP5562EVM + USB cable | M1 smoke test, M2 engine patterns, M3 pattern library |
| **B** | + NTAG5Link Click board + I2C jumper | M4 EEPROM, M5 boot-from-EEPROM, M6 SRAM mailbox |
| **C** | + Hall sensor + multimeter | M7 sleep/wake, M8 recovery, M9 power characterization |
| **D** | All Group A-C hardware | M10 LedController trait, M11 SimplePwm |
| **E** | Future | Custom PCB, companion app, NFC OTA |

## Key Principles

- **Hardware-first**: Get real LEDs blinking before designing abstractions
- **YAGNI**: No traits, formats, or protocols until we need them
- **Small milestones**: Each produces a flashable, testable result (~1 session)
- **Deferred abstraction**: LedController trait moves to M10 (after M1-M9 prove the real interface)
- **Pattern format informed by experience**: Binary format designed in M5 after M2-M3 engine programming experience

## Current Status

- Phase 0 (scaffold) complete
- M1 code ready to flash (LP5562 driver integrated, main.rs has RGBW cycle test)
- Blocked on mini-USB cable for flashing

## Reference

Full plan: [docs/DEVELOPMENT_PLAN.md](../DEVELOPMENT_PLAN.md)