v2.2.0 - Character-Aware Indexing & UTF-8 Correctness

[CrystallineCore]

Version 2.2.0

Added

• Character-aware indexing based on Unicode characters instead of raw UTF-8 bytes.
• Proper UTF-8 support for ILIKE, including case-insensitive matching on multi-byte characters.
• Improved UTF-8 decoding paths and edge-case handling.

Changed

• Switched internal index representation from byte-based to character-based semantics.
• Index traversal and lookup logic refactored to operate on logical characters.
• Improved alignment with PostgreSQL’s text comparison and planner semantics.

Fixed

• Incorrect matches caused by multi-byte UTF-8 sequences being indexed as separate entries.
• Multiple bugs affecting INSERT, UPDATE, and DELETE operations that could leave the index in an inconsistent state.
• False positives during pattern matching on non-ASCII text.
• Missed or incorrect results when using ILIKE on UTF-8 datasets.

Refactored

• Index layout and lookup internals to support character-level traversal.
• Simplified internal invariants for better maintainability and debugging.

---

This discussion was created from the release v2.2.0 - Character-Aware Indexing & UTF-8 Correctness.