v1.2.0

Version 1.2.0

Upgrade notes

• Set STORAGE_PATH in .env — required; copy from .env.example (absolute path recommended, or empty for local dev only)
• Default host is 127.0.0.1 — update .env if you previously bound to all interfaces
• Job store uses SQLite — use ./start.sh --clear-jobs (or start.bat --clear-jobs) to reset the queue after upgrade

New features

• Added timeseries/indicators support for storing indicator data, validating imports (CSV structure, values, and frequency), and querying datasets via a DuckDB-backed API
• Added optional AI metadata reviewer via ai4data (see README-reviewer.md)
• Added geospatial metadata endpoints with installation guide (see README-geospatial.md)
• Added SQLite job persistence with FIFO ordering and audit logging

Data processing improvements

• Added chunked DTA read and export for large microdata files
• Added pandas fallback for Stata files with legacy-encoded notes/labels that pyreadstat cannot decode
• Improved weight handling for integer and float fields
• Reduced CSV export memory use with threshold-gated optimized loading

Security & configuration

• Required explicit STORAGE_PATH in .env; application refuses to start if missing
• Default bind address 127.0.0.1 in start scripts, deploy scripts, and .env.example

Operations & deployment

• Improved logging with daily file rotation, timestamps, and crash handlers
• Added Linux systemd deployment guide (deploy/linux/README.md)
• Added Windows NSSM deployment guide (deploy/windows/README.md)
• Rewrote README with Conda, .venv, and install/run instructions for local and production use