Stage 0: scaffold monorepo, backend skeleton, docs, CI
Tests · Go / test (push) Successful in 32s

- go.work (Go 1.26.3) with backend module; deps added incrementally (gin+zap only)

- backend: /healthz + /readyz, env config, graceful shutdown

- docs: ARCHITECTURE, FUNCTIONAL (+ru mirror), TESTING

- PLAN.md (stage tracker + per-stage open details) and CLAUDE.md (per-stage workflow)

- .gitea go-unit CI (gofmt/vet/build/test)
This commit is contained in:
Ilia Denisov
2026-06-02 11:57:58 +02:00
commit effe6675bc
19 changed files with 1174 additions and 0 deletions
+56
View File
@@ -0,0 +1,56 @@
# Scrabble Game — Functional spec
Per-domain user stories: what each user-visible operation does. This is the
starting point for any change request that touches behaviour. The English
version is authoritative; [`FUNCTIONAL_ru.md`](FUNCTIONAL_ru.md) is a mirror for
the project owner — mirror every point edit in the same patch (translate only
the changed paragraphs). Sections deepen as stages land; *(Stage N)* marks where
the detail is authored.
## Domains
### Identity & sessions *(Stage 1 / 6)*
A player arrives from a platform (Telegram first), via email login, or as an
ephemeral guest. The gateway validates the credential once and mints a thin
session token; the backend resolves it to an internal `user_id`. Guests are
session-only with restricted features (auto-match only; no friends, stats or
history).
### Accounts, linking & merge *(Stage 1 / 10)*
First platform contact auto-provisions a durable account. From the profile a
player links additional platform identities or an email via a confirm flow;
linking an identity that already has history merges it into the current
account (stats summed, games/friends transferred).
### Lobby & matchmaking *(Stage 4)*
Bottom tab menu: **my games**, **profile**. Auto-match (always 2 players) joins
a `(variant, language)` pool; after 10 s with no human, the robot substitutes.
Friend games (24) are formed by friend list, internal ID, or deep-link.
### Playing a game *(Stage 3)*
Place tiles, pass, exchange, or resign. A play is validated against the
dictionary at submit time and scored. One hint per game reveals the best move.
The dictionary check tool is unlimited and offers a complaint. The game ends
when the bag empties and a player clears their rack, after 6 consecutive
scoreless turns, or by the 24-hour move timeout (auto-resign).
### Robot opponent *(Stage 5)*
Indistinguishable-from-human substitute in auto-match. Decides once whether to
play to win (~40%), targets a small score margin, plays with human-like timing
and a night sleep window, and nudges/answers nudges like a person.
### Social: friends, block, chat, nudge *(Stage 4)*
Add friends; block chat and/or friend requests independently; per-game chat;
nudge the awaited opponent at most once per hour (platform-native push).
### Profile & settings *(Stage 4)*
Language (en/ru), display name, linked accounts, email binding, timezone, block
toggles.
### History & statistics *(Stage 3)*
Finished games are archived in a dictionary-independent form and exportable to
GCG. Statistics: wins, losses, max points in a game, max points for one word.
### Administration *(Stage 9)*
Admin (Basic Auth at the gateway) reviews word complaints, manages dictionary
versions, and inspects users/games.