Files
dance-lessons-coach/adr/0004-interface-based-design.md
Gabriel Radureau a24b4fdb3b 📝 docs(adr): homogenize 23 ADRs + rewrite README (Tâche 7 migration) (#18)
## Summary

Homogenize all 23 ADRs to a single canonical header format, and rewrite `adr/README.md` to match the actual state of the corpus.

This is **Tâche 7** of the ARCODANGE Phase 1 migration (Claude Code → Mistral Vibe). Independent from PR #17 (Tâche 6 — restructure AGENTS.md) — both can merge in any order. No code changes; only documentation.

## Changes

### 1. Homogenize 21 ADR headers (commit `db09d0a`)

The audit (Tâche 6 Phase A, Mistral intent-router agent, 2026-05-02) had identified **3 inconsistent header formats** :

- **F1** — list bullets (`* Status:` / `* Date:` / `* Deciders:`) : 11 ADRs (0001-0008, 0011, 0014, 0023)
- **F2** — bold fields (`**Status:**` / `**Date:**` / `**Authors:**`) : 9 ADRs (0009, 0010, 0012, 0013, 0015, 0016, 0017, 0018, 0019)
- **F3** — dedicated section (`## Status\n**Value** `) : 5 ADRs (0020, 0021, 0022, 0024, 0025)

Plus mixed metadata names (Authors / Deciders / Decision Date / Implementation Date / Implementation Status / Last Updated) and decorative emojis on status values made the corpus hard to scan or template against.

**Canonical format adopted** (see `adr/README.md` for full template) :

```markdown
# NN. Title

**Status:** <Proposed | Accepted | Implemented | Partially Implemented | Approved | Rejected | Deferred | Deprecated | Superseded by ADR-NNNN>
**Date:** YYYY-MM-DD
**Authors:** Name(s)

[optional **Field:** ... lines]

## Context...
```

**Transformations applied** (via `/tmp/homogenize-adrs.py` script, 23 files scanned, 21 modified — 0010 and 0012 were already conform) :

- F1 list bullets → bold fields
- F2 cleanup : `**Deciders:**` → `**Authors:**`, strip status emojis
- F3 sections : `## Status\n**Value** ` → `**Status:** Value` (single line)
- Strip decorative emojis from `**Status:**` and `**Implementation Status:**`
- Convert `* Last Updated:` / `* Implementation Status:` / `* Decision Drivers:` / `* Decision Date:` to bold
- Date typo fix : `2024-04-XX` → `2026-04-XX` for ADRs 0018, 0019 (off-by-2-years in original)
- Normalize multiple blank lines after header (max 1)

**ADR body content is preserved unchanged.** Only headers transformed.

### 2. Rewrite `adr/README.md` (commit `d64ab02`)

Previous README had multiple inconsistencies :

- Index table listed wrong titles for ADRs 0010-0021 (looked like an aspirational forecast that never matched reality — e.g. "0011 = Trunk-Based Development" but real 0011 is absent and Trunk-Based Development is actually 0017)
- Listed entries for ADRs 0011 (validation library) and 0014 (gRPC) but **these files do not exist** in the repo
- 0024 (BDD Test Organization) was missing from the detail list
- Template still showed the obsolete F1 format (`* Status:`)
- Decorative emojis on every status entry

Rewrite :

- Index table **regenerated from actual file contents** (title from H1, status from `**Status:**` line) — emoji-free, accurate
- Notes that 0011 / 0014 are not currently in use (reserved)
- Updated template block matches the canonical format
- Status Legend extended with `Approved`, `Partially Implemented`, `Deferred`
- Added note that 0026 is the next free number for new ADRs

## Test plan

- [x] All 23 ADRs follow `**Status:**` / `**Date:**` / `**Authors:**` (verified via grep)
- [x] No more occurrences of `* Status:` (F1) or `## Status` (F3) in any ADR header
- [x] No more emojis on `**Status:**` lines
- [x] `adr/README.md` index links resolve to existing files (no more 0011 / 0014 dead links)
- [x] Pre-commit hooks pass (`go mod tidy`, `go fmt`, `swag fmt`)

## Migration context

Part of Phase 1 of the ARCODANGE migration from Claude Code to Mistral Vibe. Tâche 7 of the curriculum.

Independent from PR #17 (which restructures `AGENTS.md`). The two PRs touch disjoint files — no merge conflict expected when both are merged.

🤖 Generated with [Claude Code](https://claude.com/claude-code) (Opus 4.7, 1M context). Mistral Vibe (intent-router agent / mistral-medium-3.5) did the original audit identifying the 3 formats during Tâche 6 Phase A.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-Authored-By: Mistral Vibe (devstral-2 / mistral-medium-3.5)
Reviewed-on: #18
Co-authored-by: Gabriel Radureau <arcodange@gmail.com>
Co-committed-by: Gabriel Radureau <arcodange@gmail.com>
2026-05-03 11:01:13 +02:00

2.6 KiB

Adopt interface-based design pattern

Status: Accepted Authors: Gabriel Radureau, AI Agent Date: 2026-04-02

Context and Problem Statement

We needed to choose a design pattern for dance-lessons-coach that provides:

  • Good testability and mocking capabilities
  • Flexibility for future changes
  • Clear separation of concerns
  • Dependency injection support
  • Maintainability and readability

Decision Drivers

  • Need for easy testing and mocking
  • Desire for flexible, maintainable architecture
  • Requirement for clear component boundaries
  • Need for dependency injection
  • Long-term evolution of the codebase

Considered Options

  • Interface-based design - Define interfaces first, implement later
  • Direct implementation - Implement concrete types directly
  • Functional approach - Use functions and composition
  • DDD-style aggregates - Domain-driven design patterns

Decision Outcome

Chosen option: "Interface-based design" because it provides excellent testability, clear contracts, flexibility for future changes, and good support for dependency injection while maintaining good readability.

Pros and Cons of the Options

Interface-based design

  • Good, because excellent for testing and mocking
  • Good, because clear component contracts
  • Good, because flexible for future changes
  • Good, because supports dependency injection well
  • Good, because encourages good separation of concerns
  • Bad, because slightly more boilerplate
  • Bad, because can be over-engineered if taken too far

Direct implementation

  • Good, because simpler and more direct
  • Good, because less boilerplate
  • Bad, because harder to test and mock
  • Bad, because less flexible for changes
  • Bad, because tighter coupling

Functional approach

  • Good, because can be very clean and simple
  • Good, because good for pure functions
  • Bad, because less familiar in Go ecosystem
  • Bad, because harder to manage state

DDD-style aggregates

  • Good, because good for complex domains
  • Good, because clear boundaries
  • Bad, because overkill for simple services
  • Bad, because more complex to implement

Implementation Examples

// Good: Interface defined first
type Greeter interface {
    Greet(ctx context.Context, name string) string
}

type Service struct{}

func (s *Service) Greet(ctx context.Context, name string) string {
    // implementation
}

// Bad: Direct implementation without interface
type Service struct{}

func (s *Service) Greet(name string) string {
    // implementation
}