What's fixed
Hotfix release for two reliability issues, both reported within hours by real users - thank you! 🙏
- 🚀 No more "stuck at Loading speech model…" two causes fixed: an interrupted
first download left a partial model cache that made startup silently re-download while
claiming to "load", and a slow/blocked connection to huggingface.co could hang the
model load even when it was fully on disk. WritHer now validates the cache properly
and, when the model is complete, loads it with zero network access startup is
also noticeably faster (~1s model load instead of several seconds). - 👻 Ghost process, third and final round - if Tk failed internally while closing
(e.g. quitting right after using Settings), the process could still survive the quit.
A watchdog now force-terminates the process 5 seconds after quitting, no matter what.
If quitting v1.4.0 ever leaves WritHer "already running" with no tray icon:
Task Manager → end WritHer.exe, then upgrade. - ⌨️ Bare-modifier hotkeys (e.g. Right Ctrl alone) configured in config.py now work
instead of being silently ignored (#19, thanks @MaxiKingXXL).
Full Changelog: v1.4.0...v1.4.1
Contributors
MaxiKingXXL