Releases: Aschefr/Alanbix
Releases · Aschefr/Alanbix
Release list
v1.32.1 - Fix connexion Docker & Unraid (Port 41481)
🐛 Correction — Connexion Docker Standalone & Unraid
- Rétablissement de la connexion conteneur (Port 41481) : Correction du routage des requêtes API et WebSockets en production. Le conteneur unifié utilise désormais fidèlement le port unique d'Alanbix (41481) ou votre reverse proxy sans tenter de contacter le port de développement 8000.
v1.32.0 - Chat public moderne, anti-lag LAN, harmonisation visuelle & calcul des points
💬 Chat Public Moderne & Interactif
- Réactions Émojis en direct : Réagissez aux messages avec un sélecteur d'émojis complet (catégories LAN/Gaming, recherche instantanée) et un accès direct aux 4 émojis récents.
- Réponses & Citations contextuelles : Répondez précisément à un message avec aperçu au-dessus de la saisie et clic sur la citation pour retrouver le message d'origine dans le fil.
- Annonces Épinglées : Les organisateurs peuvent épingler des messages importants, visibles sous forme de bannière en haut du chat avec accès direct au message.
- Indicateurs de saisie en temps réel : Visibilité immédiate des personnes en train d'écrire, avec des punchlines humoristiques et le statut de réflexion du bot
@Alanbix. - Alertes sonores douces : Signal audio léger lors d'une mention directe
@votre_nom(désactivable d'un clic avec le bouton silencieux).
⚡ Fluidité & Performances en LAN (Anti-Lag)
- Protection anti-saturation (Thundering Herd) : Fin des ralentissements réseau lors des saisies simultanées de scores ou des lancements de tournois grâce au lissage intelligent des actualisations.
- Défilement ultra-fluide du Chat : Chargement progressif des anciens messages à la demande (défilement continu sans saut d'écran) et affichage instantané même en cas de spam d'émojis ou de GIFs.
- Base de données accélérée : Indexation complète pour des temps de réponse instantanés sur l'ensemble du tableau de bord.
🎨 Design, Lisibilité & Cohérence Visuelle
- Harmonisation complète des thèmes (Sombre & Clair) : Contrastes, typographies et couleurs de fond ajustés sur toutes les pages (Règlement/Info, Profil, Arène Spectateur, Paramètres Admin).
- Modales & Fenêtres unifiées : Nouveau design standardisé pour toutes les boîtes de dialogue et popups avec typographie moderne (Outfit) et fermeture rapide par la touche
Échap. - Studio Avatar amélioré : Grille de transparence visible pour faciliter le détourage et l'ajustement des photos de profil.
🏆 Calculs & Administration
- Affichage propre des points : Élimination définitive des artefacts de calcul décimal (ex.
24.7 ptsnet au lieu de24.700000000000003 pts). - Nouveau panneau d'administration du Chat : Contrôles réorganisés en cartes claires, avec sauvegarde instantanée des réglages anti-spam et du slowmode.
- Traductions complètes (i18n) : 100 % de l'interface traduite en français, anglais et espagnol sans texte tronqué ni bouton brut.
v1.31.0
v1.30.2
Features & Bug Fixes — RAG Vectorization, LLM Rate Limit Context, and Notifications Sync
- RAG Auto-Vectorization: Modified approve_rag_suggestion to lookup the correct active Ollama instance dynamically instead of using localhost default. Added real-time refresh sync on Svelte UI using knowledge_updated WS events.
- LLM Rate Limit Context Injection: Added system prompt restrictions in stream_query detailing active call_admin cooldown/limits, allowing local and remote LLM models to know rate limits beforehand instead of generating false promises or contradicting themselves.
- WebSocket Broadcast Simplification: Simplified _tool_call_admin and _tool_suggest_rag_entry WebSocket broadcast tasks to use a single global 'notification_new' event, avoiding database session context expiration (DetachedInstanceError).
- Notification Details Navigation: Added Action detail redirect buttons on RAG notifications, routing administrators directly to the Conversations IA tab (tab=conversations) and automatically scrolling to target elements.
v1.30.1
v1.30.0
Features & Security — AI Admin Calls, RAG suggestions, and Real-Time Chat Sync
- AI Admin Calls Tool & Manual Button: Implemented the
call_admintool for the AI assistant and a manual "Call Admin" button for players. Added configurable anti-spam limits (cooldown period, daily user limit, global hourly limit) to prevent spam. - Separated Call Admin Configuration: Split the configuration toggles so that the manual button and the AI tool calling capability can be enabled or disabled independently from the admin panel.
- RAG Knowledge Suggestion panel: Created a new workflow where the AI assistant can suggest additions to the RAG knowledge base. Admins can review, edit, approve, or reject suggestions directly from a newly designed "Conversations IA" section in the admin panel.
- Multi-language Error Handling: Modified backend cooldown verification endpoints to return structured data (
error_codeandparams) instead of hardcoded French text. Enhanced the client API class to pass structured errors, enabling dynamic translation of cooldown limits using$tSvelte helper on the player's view (supporting French, English, and Spanish). - Notification Filtering via Active Chat Tracking: Introduced an
/active-chat/{conv_id}endpoint and a frontend ping interval (every 4 seconds) to track which conversation a player is actively viewing. If the player is looking at the chat window, database notifications are skipped on admin message interventions to prevent unread notification badges from flashing. - Real-Time WebSocket Message Deletion Sync: Integrated WebSocket broadcasts (
ia_message_deleted) when messages are deleted. Implemented local action locks (isProcessingLocalAction) on the frontend to avoid race conditions and infinite reload loops during batch message deletions. - AI Response Regeneration Fix: Resolved a frontend bug where the "Retry" button on bot messages blocked the request by setting
loading = truetoo early. Added support forskip_save_userpayload flag to prevent prompt duplication in both the database and the Svelte UI during regeneration. - Unified Button Styling: Replaced the default unstyled
btn-ghostwith the project-wide semi-transparentbtn-secondaryclass for Dismiss (Admin calls) and Reject (RAG suggestions) actions.
v1.29.1
v1.29.0
v1.28.0
Features & Improvements
- OpenAI API Support: Full compatibility with OpenAI-compatible backends like LM Studio, llama-server, KoboldCPP, etc.
- Auto-detection: Automatically routes API payloads and parses SSE stream chunks depending on the detected endpoint protocol.
- I18n & Migrations: Refreshed Ollama/OpenAI labels and added startup overrides cleanup.
v1.27.1 - Room Plan i18n Fix & English Changelog
Features & UI — Room Plan and Dashboard UI Improvements
- Tournament Tabs Stability: Fixed a visual regression where active tournament tabs (.running-tabs) in the dashboard could disappear vertically upon click. Added \lex-shrink: 0\ and a minimum height to stabilize the layout.
- Leaderboard Avatar Protection: Prevented visual compression of avatars (.lb-avatar) and rank badges (.lb-rank) by very long usernames in the dashboard leaderboard by forcing \lex-shrink: 0. Long usernames are properly truncated with an ellipsis.
- Team Names in Welcome Wizard: Added dynamic and scaled (\ extLength\ / \lengthAdjust) team name rendering on occupied seats in the simplified room plan shown during the welcome wizard.
- Teammate Seat Coloring: Implemented a distinctive orange/amber color scheme (--map-seat-teammate-fill\ / --map-seat-teammate-stroke) for seats occupied by teammates across all room plan views (full map, dashboard mini-map, and welcome wizard). The map legend dynamically updates to include teammates if the user has a team.
- Teammates i18n Keys: Added and synchronized the \map_legend_teammates\ translation key in \r.json, \en.json, and \es.json\ (preserving UTF-8-sig encoding) for correct teammate seat rendering labels.