Fixes & Enhancements
* Resolved five critical Apprise notification issues:
• Ensured configuration reload during scheduled jobs
• Fixed warranty data fetching for Apprise-only users
• Refactored notification dispatch logic with dedicated helpers
• Corrected handler scoping via Flask app context
• Wrapped scheduler jobs with Flask app context to prevent context errors
→ Verified: Scheduled Apprise notifications now work reliably for "Apprise only" and "Both" channels.
* Added support for SMTP\_FROM\_ADDRESS environment variable, allowing sender address customization independent of SMTP username. (PR #115)
* Fixed duplicate scheduled notifications in multi-worker environments:
• Strengthened should\_run\_scheduler() logic
• Now guarantees exactly one scheduler instance across all Gunicorn modes.
* Fixed stale database connection handling in scheduled jobs:
• Fresh connection acquired each run, properly released via try/finally
• Eliminates "server closed the connection" errors.
* Definitive scheduler logic fix for all memory modes (ultra-light, optimized, performance):
• Single-worker runs scheduler if GUNICORN\_WORKER\_ID is unset
• Multi-worker: only worker 0 runs scheduler.
Impact
* Apprise and Email notifications are now stable, reliable, and production-ready
* No more duplicate or missed notifications across all memory modes
* Improved system efficiency and robustness
### Added
- Collapsible warranty details in the Status page table, showing full warranty info inline.
- Inline warranty editing via "Edit Warranty" button in expanded view.
- Vendor column added to sort options on the main warranty list.
- Vendor filter dropdown added to the main warranty list.
### Changed
- Removed direct product name links from Status page table in favor of inline expansion.
- CSV export now formats dates as YYYY-MM-DD.
### Fixed
- Resolved browser validation error with product URLs in hidden tabs by switching to text input and improving validation logic.
- Edit modal no longer closes on backdrop clicks.
- Migration scripts now support custom `DB_NAME` values by removing hardcoded names.
