- Persist display names in per-channel JSON cache files (Name +
DisplayName fields on cachedChannel). GetAllTitles reads all ch_*.json
files; PutTitle updates a channel file in-place without losing messages.
- Replace blocking FetchTitles calls in fetchMeta and refreshChannel with
an instant disk read (GetAllTitles) applied before the SSE broadcast,
so channels appear with cached titles immediately on every load.
- ensureTitlesFetched runs in a single background goroutine (titlesMu +
titlesLoading guard prevents duplicates). On success it persists titles
and pushes an SSE update. On error or empty response it backs off for
5 minutes so an old server does not cause endless retries.
- Block 0 of TitlesChannel now carries a uint16 total-block-count prefix
(added in rebuildTitlesBlocks). FetchTitles reads the count from block 0
and fetches all remaining blocks in parallel instead of sequentially.
- FetchTitles timeout raised from 10 s to 1 minute.
- Add dedicated TitlesChannel (0xFFF9) following the VersionChannel pattern
- Server encodes name→title map via EncodeTitlesData/DecodeTitlesData
- Metadata wire format unchanged for backward compatibility with old clients
- All three fetchers (public Telegram, MTProto, X/Nitter) extract and store display names
- Client fetches TitlesChannel with a 10s deadline; falls back to channel handles gracefully on old servers
- Old clients are unaffected — they never query 0xFFF9
- Implemented /api/cache/clear endpoint to delete all files in the cache directory.
- Added tests for cache clear functionality, including cases for empty cache and method not allowed.
- Introduced /api/rescan endpoint to trigger a manual rescan of the configured channels.
- Added tests for rescan functionality, covering cases for not configured, method not allowed, and successful rescans.
- Enhanced server initialization to bootstrap configuration from active profile if config.json is missing.
- Refactor fetcher tests to improve coverage and add new tests for resolver scoring and query handling.
- Update ResolverChecker to retry every minute until a healthy resolver is found and adjust timeout settings.
- Introduce scatter parameter in the configuration to allow concurrent DNS requests, with UI adjustments for user input.
- Modify metadata fetching logic to utilize cached data efficiently and ensure fresh data is fetched when necessary.
- Implement server-side handling for fetch progress, including countdowns and UI updates during long fetch operations.
- Added Profile and ProfileList structs to manage user profiles with unique IDs and nicknames.
- Introduced endpoints for CRUD operations on profiles: `/api/profiles` for managing profiles and `/api/profiles/switch` for switching active profiles.
- Implemented settings management with an endpoint `/api/settings` to handle user preferences like font size and debug mode.
- Enhanced the server to load and save profiles from a `profiles.json` file.
- Updated the fetcher initialization to respect the active profile's configuration.
- Added comprehensive end-to-end tests for profiles and settings APIs to ensure functionality and persistence.
Sarto