v4.0.0

Breaking changes:

• Index format 4 — existing indexes must be re-scanned
• sist2-admin rewritten in Node.js; jobs and schedules are not migrated (sorry!)
• Name entity recognition removed

New features:

• Thread pool reworked: libuv for inter-process communication.
• .doc extraction moved to libantiword2 - should fix most of the crashes / memory leaks related to MS Word files.
• PDF parsing performance improvements, 2x-3x speedup.
• Reworked build process; releases now come with fully static binaries.

Bug fixes:

• Incremental scans keep documents nested in unchanged archives.
• Bumped up dependency versions.

Full Changelog: 3.4.7...v4.0.0