- Add byse.sx health check via API upload/server endpoint
- Virtual scrolling for queue table (>200 rows renders only visible rows)
- O(1) job lookups via index Maps instead of O(n) array.find()
- Event delegation on queue tbody instead of per-row listeners
- Async config writes to avoid blocking main process
- Increase persist debounce to 3s during uploads (was 250ms)
- Reduce debug logging to state changes only
- Move save button to bottom-right in settings
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Root cause: startBatch() ran synchronously inside ipcMain.handle()
callback, causing webContents.send() events to conflict with the
handle response and never reach the renderer.
Fix: defer startBatch() via process.nextTick so IPC response is
sent first, then upload events flow correctly.
Also:
- Add .catch() on startBatch to surface hidden errors
- Fix settings panel not updating after save (renderSettings)
- Add select-folder IPC handler (was in preload but missing)
- Add debug-log and debug-test-upload IPC for diagnostics
- Add upload-debug.log file for tracing upload flow
- Add unhandledRejection handler for main process
- Add scramble defaults to config-store globalSettings
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add max redirect depth (10) to Vidmoly _fetch to prevent stack overflow
- Drain undici response body on redirect to prevent connection leaks
- Fix installUpdate unhandled promise rejection in main.js
- Fix context menu submenu viewport overflow with flip-left CSS
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- updater: replace undici.request() with fetch() (fixes maxRedirections
error that blocked auto-update from v1.0.0 to v1.1.0)
- upload-manager: move signalCleanup declaration outside try block
(was causing ReferenceError in catch, silently breaking ALL retries)
- upload-manager: _combineSignals now returns cleanup fn to prevent
abort listener accumulation over batch lifetime
- upload-manager: _sleep removes abort listener on normal timer fire
- hosters: apiGet removes abort listener in finally block
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Semaphore.acquire() now accepts AbortSignal — waiting jobs are properly
removed from queue on abort, preventing startBatch from hanging forever
- Clamp upload progress to 0-100% in both upload-manager and renderer
- Upload-manager handles semaphore abort rejection gracefully
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add FIFO semaphore for per-hoster concurrency control
- Add token-bucket speed limiter with abort signal support
- Rewrite upload-manager with retry loop, speed monitoring, and rich progress events
- Add per-hoster settings: retries, max speed, parallel count, restart below speed, time interval, max size
- Add context menu with shutdown-after-finish (sleep/shutdown/restart), always-on-top
- Add z-o-o-m-style queue table with 8 columns, status-colored rows, progress bars
- Add debounced queue rendering with scroll position preservation
- Add statusbar with global speed, total bytes, elapsed time
- Fix speedMonitor interval leak on error and scoping bug
- Fix throttle not respecting abort signal during cancellation
- Fix combined signal listener cleanup
- Bump version to 1.1.0
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
