- GET_VIDEO_STATE now returns allVideos[] with summary per video element
- Copy report shows markdown table when multiple videos on page
- Table includes: resolution, muted, paused, readyState, duration, and marks selected target
- Helps diagnose Prime Video scenarios where preview vs main video both exist
- findVideo() now scores all video elements by size + unmuted + duration
- Fixes Prime Video selecting 0x0 placeholder instead of actual player
- Fix history entries showing '?' by using h.action instead of h.event/h.type
- Update Prime Video status in TESTED_SERVICES.md to partial support
- Copy Logs button now generates full Markdown debug report (system, tab, connection, video, history, logs)
- Enhanced Dev tab with 20+ video state fields (network, dimensions, error, shadow DOM, platform detection)
- Platform auto-detection (YouTube, Netflix, Twitch, Prime Video, Disney+, Hulu, HBO, Vimeo, Dailymotion)
- 'No video found' diagnostic mode with hints
- Convert mockup from inline <span lang> to {{MOCK_*}} template vars (all 6 languages)
- New TwoPointZero branding: icons (16/32/48/96/128), favicons (PNG 16/32 + apple-touch + 192)
- Fix logo vertical centering via display:contents on picture wrappers
- Increase popup logo to 48px, website nav logo to 64px
- AVIF quality 70->80, speed 6->4, remove min 3KB threshold
- Add 32+96 icon sizes to manifest
- Remove amazon. from tab blacklist (unblocks Prime Video)
- Update TESTED_SERVICES.md: mark Prime Video as non-functional
- Update CONTRIBUTING.md with Copy Debug Report workflow
- Add currentServerUrl tracking: only disconnect/reconnect when server URL actually
changes, otherwise reuse existing connection and just switch rooms
- Add resolveServerUrl() helper for consistent URL resolution across handlers
- forceDisconnect() now resets currentServerUrl to null
- CONNECT and WEB_JOIN_REQUEST handlers: smart comparison before reconnecting
- Add generateRoomId(): human-readable 'ADJECTIVE-NOUN-Number' format (e.g. HAPPY-KOALA-16)
- Replace Math.random() room IDs with generateRoomId() in joinBtn and handleCreateRoom()
- Expand USERNAME_ADJECTIVES: +16 words (Turbo, Zen, Pixel, Cyber, Solar, etc.)
- Expand USERNAME_NOUNS: +12 words (Yeti, Goblin, Pirate, Ninja, Wizard, Storm, etc.)
- Add emoji mappings for all new words
- WEB_JOIN_REQUEST handler: forceDisconnect+connect instead of reusing old socket
(same bug as CONNECT: JOIN_ROOM was sent to official server instead of custom)
- joinBtn handler: generate random password when roomId is empty and password empty
(previously created rooms without passwords when using the advanced join button)
- Auto-copy invite link when creating room via joinBtn (consistent with Create Room button)
- Fix: CONNECT/RETRY_CONNECT always force-disconnect before reconnecting
(prevents JOIN_ROOM going to official server when custom is selected)
- Add forceDisconnect() helper: clears socket, eventQueue, episodeLobby,
expectedAcksCount, broadcast status, persists cleanup to storage
- Fix: save useCustomServer to storage on Join Room click (no race condition)
- Fix: trigger RETRY_CONNECT on server mode toggle and custom URL change
- Fix: show error when Custom selected but no URL entered (no silent fallback)
- Fix: button label changed to 'Join / Create Room' in all 6 locales
- Rewrite scheduleReconnect() for two-phase strategy:
Phase 1 (aggressive): 500ms-5s backoff, max 20 attempts or 5 minutes
Phase 2 (slow): retry every 5 minutes indefinitely, never give up
- Persist reconnectAttempts/reconnectStartTime/reconnectFailed to session storage
- Remove 'reconnect_failed' status; slow mode shows as 'reconnecting' with retry button
- Add scheduleReconnect() call on offline path to continue retry cycle
- Add sr-only CSS class for screen-reader/crawler-visible elements
- Add hidden <nav> with real <a> links to all 6 language versions in template.html
- Googlebot can now discover /de/, /fr/, /es/, /pt-BR/, /ru/ via HTML links
- No visual changes - existing <select> dropdown remains the user-facing switcher
- Rebuild all www/ output with updated CSS hash
HowTo schema was hardcoded English — now uses {{HOWTO_*}}
placeholders. All 3 rich-result schemas are now locale-aware:
- FAQPage (expandable SERP snippets)
- HowTo (step-by-step SERP snippets)
- SoftwareApplication (entity understanding)
SoftwareApplication and Organization use language-neutral
brand name — no locale keys needed.
- lastmod bumped to 2026-06-01 for all 6 language roots
- xhtml:link hreflang entries added per URL (Google-recommended
belt-and-suspenders approach alongside HTML hreflang tags)
- FAQPage schema now uses {{FAQ_Q1}}..{{FAQ_A6}} placeholders
instead of hardcoded English — each language gets its own
schema for locale-matched Google Rich Snippets
- Q6 updated across all 6 languages: 'What if I don't trust
the official server?' + self-hosting Docker mention
PageSpeed Insights flagged AVIF files as uncached because the
@static matcher only included *.webp. Adding *.avif enables
1-year Cache-Control for all AVIF assets.
Google does not support sameAs for app store links in SoftwareApplication
rich results. sameAs is for identity pages only (Wikipedia, Wikidata, GitHub).
Store pages are independently indexed by Google and need no schema linking.
Google treats SoftwareApplication.sameAs as app listing references
(not social profiles like Organization.sameAs). This connects
KoalaSync's entity graph to both browser stores without
duplication risk.
- Add LCP image preload for hero mascot
- Add HowTo structured data schema (3-step install guide)
- Update SoftwareApplication schema with screenshot + image fields
- Add glassmorphic :focus-visible outlines for keyboard accessibility
- Add content-visibility: auto to heavy offscreen sections for INP
- Expand lang-init.js auto-redirect to all 6 supported languages
- Remove impressum/datenschutz from sitemap.xml (noindex conflict)
- Removed unused html variable from localizeHomeLinks in website/app.js.
- Added website/build.js to ESLint flat config Node files match pattern to define require, process, and __dirname.
- Replaced /impressum.html and /datenschutz.html with clean URLs /impressum and /datenschutz in sitemap.xml.
- Recompiled website to copy clean sitemap to www/ folder.
- Restructured translation guide with clean markdown layout and tables.
- Removed mermaid sequence diagram to improve text-based clarity.
- Audited steps for contributing new language keys.
- Corrected static server root path to website/www/ in Caddyfile.example and website/README.md.
- Updated core roles to reflect multi-language (EN/DE/FR/ES/PT-BR/RU) support.
- Refactored local development and compilation instructions to target build.js and www/ output.
- Added pure Node.js dynamic i18n static site generator (build.js).
- Structured locales for English, German, French, Spanish, Brazilian Portuguese, and Russian.
- Replaced two-state toggle with premium glassmorphic language select dropdown.
- Integrated robust segment-based locale routing with safe dynamic fallbacks for legal and invite pages.
- Audited Core Web Vitals (LCP preloads, CLS dimensions) and SEO structures (robots, sitemap).
- Added dedicated Localization section to README and created contributor TRANSLATION guide.
Timo