Skip to content

ROADMAP.md

Latest commit

 

History

History
74 lines (53 loc) · 3.21 KB

ROADMAP.md

File metadata and controls

74 lines (53 loc) · 3.21 KB
Raw

Roadmap

Four phases. Each builds on the last; the data model and module boundaries are already designed so later phases are additive (no rewrites). For day-to-day state see PROGRESS.md.

Phase 1 — MVP: a fully usable typing game (no database needed to play)

Adds: event loop, pure engine, type-along render, themes, figlet banners, config, random + PDF/DOCX challenge sources, timed + word-count presets.

Scope

  • Event-driven loop (poll(timeout)+read(), filter KeyEventKind::Press, dirty-flag redraw — no fixed 60 fps tick).
  • RAII teardown + panic hook; kitty enhancement flags gated behind supports_keyboard_enhancement().
  • Pure engine: Type / Backspace / DeleteWord / NextWord, grapheme-indexed, pure metrics.
  • Per-character colored render (cache, mutate only changed states); figlet timer/WPM; results screen.
  • Config + ≥2 themes + a 16-color fallback, from the XDG config dir with embedded defaults.
  • Sources: random wordlist + import PDF/DOCX → normalizer → passage.

Acceptance criteria

  • type-cli --time 60 runs a 60 s test end-to-end with live WPM/timer, correct/incorrect coloring, and a results screen (net + raw WPM, accuracy, consistency).
  • type-cli import file.pdf and file.docx yield a typeable normalized passage; a scanned/no-text PDF shows a clear message.
  • No terminal corruption on panic or quit (RAII teardown verified).
  • Engine, stats and the normalizer are unit-tested and green.

Phase 2 — Persistence, stats & charts

Adds: SQLite (rusqlite bundled) + migrations (rusqlite_migration), run history, charts, most-failed-key detection, worst-word retry.

Scope

  • Persist each run + char_stat + worst_word after a test; WAL pragmas; idempotent migrations.
  • Results Chart (WPM line + accuracy/area), BarChart of most-failed keys.
  • type-cli stats: WPM/accuracy over time, daily buckets, keyboard heatmap (colored Spans over QWERTY).
  • "Retry worst words" → starts a new session from a run's worst words.

Acceptance criteria

  • Runs persist across launches; type-cli stats renders history + most-failed key (min-sample gated).
  • Retry-worst-words works; migrations run cleanly from an empty DB.

Phase 3 — Ghost / shadow replay

Adds: keystroke-timeline capture + race-against-a-prior-run.

Scope

  • Capture (t_offset_ms, action) (row table while prototyping → compressed BLOB in production).
  • Ghost mode replays a prior run through the same pure engine (clock-as-parameter pays off); render the shadow cursor; record ghost_of_run_id.

Acceptance criteria

  • "Race your best 60 s run" → the ghost advances on its recorded schedule; the result records which run was raced; replay is deterministic.

Phase 4 — Online / multiplayer

Adds: identity sync + networked sessions.

Scope

  • Populate remote_id / synced_at; upload unsynced runs.
  • A network thread feeds events via mpsc; the loop selects over input + network using the same Action path. Backend protocol/auth/hosting decided in a dedicated P4 design spike (future ADR).

Acceptance criteria

  • Local results upload; race vs a remote ghost / leaderboard; offline-first preserved.
  • All additive — no schema rewrites.