This release focuses on preventing user interface freezing, optimizing memory usage during configuration updates, and eliminating redundant code structures to ensure a smooth user experience.
What’s New:
- Non-Blocking Process Cleanup (UI Freeze Fix): Migrated the process cleanup and port polling operations inside
server.pyto a backgroundQThreadcombined with a localQEventLoop. This ensures the main user interface remains responsive and never enters a “Not Responding” state while starting or stopping the server. - Memory & Instance Optimization: Updated
CommandPreviewWidgetto use a shared reference of the mainLlamaServerManagerinstance instead of instantiating a new manager on every single form input modification. This drastically reduces redundant CPU and memory overhead during configuration tweaks. - Safe Initialization Sequence: Reordered the initialization steps inside
ui.pyinto a strict 8-step sequence. Core properties and logging callbacks are now guaranteed to be fully initialized before any signals are connected or configurations are loaded. - Dead Code Elimination: Removed the unused
_continue_start_serverslot and redundant signal connections fromserver.py, keeping the codebase cleaner and more maintainable.