Updates
- Custom tags delimiters are now always applied if set (before: only if the file does not contain any multi-valued tag)
- Added a way to dump internal execution traces, to better understand performance issues (to be enabled via
tracing-levelinlms.conf) - Various performance improvements:
- File scanning jobs are distributed more efficiently between threads
- The last stage of the analysis (checking the existence of files) is way faster on very large databases (>500k files)
- Removed some useless writes in the database