Oyma v0.9.0: Local SQLite FTS5 Search & Query Engine

Release Highlights: Lightning-Fast Vault Retrieval

As personal knowledge bases expand to thousands of daily notes, meeting transcripts, and project specifications, finding exact information quickly becomes critical. Version 0.9.0 introduces our embedded SQLite FTS5 search engine.

Architectural Improvements

Embedded SQLite FTS5 Indexing

Instead of loading entire document trees into JavaScript heap memory, Oyma maintains an incremental SQLite database containing a virtual full-text search table (fts5). Whenever a Markdown file is created or edited on disk, an OS file watcher updates the token index in milliseconds.

BM25 Ranking & Exact Phrase Matching

Search queries are ranked using the BM25 probabilistic relevance algorithm, prioritizing notes where keywords appear in headings or frontmatter titles. Users can search with exact quotes ("ScreenCaptureKit") or boolean exclusions (audio NOT cloud).

Storage Footprint & Index Performance

We conducted rigorous stress testing on an archival test vault containing 50,000 Markdown documents:

• Initial indexing completed in 3.8 seconds on Apple Silicon SSD storage.
• The incremental SQLite index database occupied less than 2.8% of the raw document file size.
• Average query execution time was measured at 4.2ms for single keywords and 8.7ms for complex boolean queries.

Explore our full-text search feature guide, learn how to use the command palette, calculate document length with the markdown word counter, and read about Obsidian vault compatibility.