Align the OpenUP markdown set to the blank templates (process-data repo, templates/openup/) per BUILD-CHECKLIST 7. Scenario step text unchanged; the EA model needs no change (brief descriptions already live in the model's UC/actor notes). - vision.md: positioning tables, stakeholder summary, user environment, needs-and-features table (R1/R2), other product requirements - use-case-model.md: per-UC specification sections (brief description, actor descriptions, named alternative/exception subsections with numbered steps, post-conditions, special requirements) - supplementary-specifications.md: category tables with priority and use-case columns + reverse coverage matrix; IDs unchanged - test/quality-check-procedure.md: migrated from Joplin (single content home); stale open items marked resolved (WI-1/WI-2) - analysis-design/sequences/: 12 PlantUML design sequences (lifelines = analysis classes, messages = design elements; per the PlantUML layout style guide) + PNG renders; embedded in Architecture Notebook 9.3 - Joplin consolidated: project content notes deleted (all live in docs/game + docs/openup); Joplin keeps the status note only
4.6 KiB
4.6 KiB
Quality Check Procedure
Skara Brae — OpenUP Test · migrated 2026-10-05 from the Joplin note of the same name
(markdown is the single content home; Joplin keeps status only). Post-migration update:
the three unchecked items below were resolved — rand::rng() direct use by WI-1
(single construction point request_rng()), handler-side business logic by WI-2
(run_adventure extraction) and constants by WI-2 (sb-core::constants).
Purpose
Systematic verification that the implementation matches design notes, passes all code checks, and follows Clean Code / Clean Architecture principles.
Procedure
Step 1: Code Checks
# Rust build + clippy
cd ~/Development/skara-brae/skara-brae-server && cargo clippy -- -W clippy::all 2>&1
# Rust tests
cd ~/Development/skara-brae/skara-brae-server && cargo test 2>&1
# Vue typecheck
cd ~/Development/skara-brae/skara-brae-client && npx vue-tsc --noEmit 2>&1
Step 2: Implementation vs Design Notes
For each design note, verify the implementation matches:
| Note (docs/) | What to Check |
|---|---|
docs/game/game-concept-overview.md |
Character system (races, classes, stats), economy model, adventure structure |
docs/game/combat-system-stat-scaling-design.md |
BIGINT stats, open-ended d100, crit tables, crit caps by level diff |
docs/game/mechanics-source-mapping.md |
BT1/RM/IF/Torn mapping consistency |
docs/game/tech-stack-architecture.md |
Crate structure, dependency direction, deployment |
docs/game/skill-magic-system.md |
Spell system design vs implementation |
docs/game/medals-honours-system.md |
Medal system (not yet implemented) |
docs/game/idle-fantasy-inspirations.md |
Features adopted from IF |
docs/openup/analysis-design/architecture-notebook.md |
Layers, dependency rule, key abstractions |
Check items:
- Stats are i64 (BIGINT) — not i32
- 5 races match design (Human, Elf, Dwarf, Hobbit, HalfElf)
- 10 classes match design (Warrior through Sorcerer, no Cleric)
- Crit damage cap by level difference implemented
- Open-ended d100 combat rolls
- BT1 items (127), enemies (125), spells (79), songs (6) all present
- Zones match BT1 dungeon structure (8 zones)
- Shop economy uses BT1 gold values
- Bard songs with combat integration
- Spell system with SP costs, learning, combat casting
- Training costs gold (not just energy)
- Review Board level-up: XP threshold polynomial (18×100×(level-1)²), gold cost, max level 100
Step 3: Clean Architecture Audit
Check dependency direction:
sb-shared→ no dependencies on other game crates (domain types only)sb-core→ depends only onsb-shared+sb-common(domain services, no HTTP)sb-api→ depends onsb-core,sb-shared,sb-auth,sb-db(HTTP adapters)
Check layer violations:
- No
serde_jsonin domain logic (only in LazyLock data loaders) - No
include_str!called during business logic (only module-level const) - No
rand::rng()in domain functions — resolved by WI-1: single construction pointrequest_rng(), domain takes injected rng - Route handlers delegate to domain services — resolved by WI-2 for adventure; remaining handler-side logic (shop, heal, parts of spells) is recorded debt (Architecture Notebook D1, follow-up question Q3)
Check code quality:
- No magic numbers — resolved by WI-2:
sb-core::constantsis the audit home - No abbreviations in variable names
- Functions do one thing, are short
- No duplicated logic across modules
- No comments that restate code
- Fail fast, return early — happy path flat
- No mixed concerns (HTTP + business logic in same function)
Step 4: Bug Scan
- All gold/energy/nerve/SP costs are both displayed AND enforced
- All state mutations happen inside write locks
- No unwrap() on fallible operations in request paths — WI-2 §6 ledger classifies the remaining ones as documented keep-with-reason invariants
- All enum match arms exhaustive
- No integer overflow paths (xp_for_level uses saturating arithmetic)
Audit History
| Date | Clippy | Tests | vue-tsc | Criticals Found | Criticals Fixed |
|---|---|---|---|---|---|
| 2026-08-03 #1 | 18 warnings | 0 tests | clean | 3 | 3 |
| 2026-08-03 #2 | 0 | 42 pass | clean | 4 | 4 |
| 2026-08-03 #3 | 0 | 44 pass | clean | 1 | 1 |
Quality Gate
A check passes when:
- Zero clippy warnings (excluding allowed lints)
- Zero typecheck errors
- All tests pass
- Zero critical architecture violations
- All tracked inconsistencies have a ticket or accepted rationale