Logging and Diagnostics
Kawaikara initializes one electron-log pipeline at the beginning of the Main process. Existing console.debug/info/warn/error calls are routed through that pipeline, so managers and bundled plugins are persisted without each subsystem configuring its own file transport. LoggingManager owns initialization, runtime level changes, renderer attachment, diagnostic-directory access, and session shutdown. It is registered in the application manager container and injected into consumers that need a scoped logger or renderer capture. New code can use logging.createLogger(scope) from src/Main/Manager/LoggingManager.ts when a stable subsystem label is useful.
Output and retention
The active file is UserRoot/KawaiData/logs/kawaikara.log. Electron’s own
userData and sessionData paths point to UserRoot/Electron, keeping runtime
caches separate from Kawaikara-owned settings, library state, and diagnostics.
It rotates at 5 MiB using electron-log’s archived predecessor file. Development
builds record debug and above; packaged builds record info and above. A
runtime KAWAIKARA_LOG_LEVEL value can select error, warn, info,
verbose, debug, or silly.
Each process launch records a random session ID, app/runtime versions, release channel, platform, architecture, renderer mode, startup completion, and clean shutdown. Preferences > App Info > Diagnostic logs opens the containing folder.
Captured failures
- Unhandled Main-process errors and promise rejections.
- Electron child/renderer process termination, preload failures, and navigation load failures.
- Console output and unresponsive/responsive transitions from app-owned Viewer, Overlay, Video, and PiP surfaces.
- Existing manager, site, updater, native Video, and HLS error messages.
Remote sites do not receive a logging bridge. Their JavaScript console chatter is not copied into the file, although Electron-level load and process failures remain visible.
Privacy boundary
Before transport, the logger recursively redacts common authorization, cookie, password, API-key, credential, and token fields. HTTP(S) URL credentials, queries, and fragments are removed. Logs remain local and are never uploaded by the application. Contributors should still avoid logging media contents, cookie jars, request bodies, or account data.