v1.0.0: Initial Release - The SIMD-Accelerated Search Engine

First stable release of Lumina Search (v1.0.0)! 🚀

Lumina is an ultra-fast, modern full-text search engine built from scratch using the latest features of Java 22. It challenges traditional solutions by ditching legacy abstractions in favor of direct memory access and CPU-level hardware vectorization.

🔥 Key Highlights in v1.0.0:

• SIMD-Powered BM25 Scoring: Relevance math is vectorized using the Java Vector API (jdk.incubator.vector), processing 256-bit blocks of documents in a single CPU cycle.
• Zero-Copy Memory Access: Index and dictionary files are memory-mapped directly from disk using the FFM API (MemorySegment and Arena), bypassing the Java Heap and keeping the GC idle.
• Virtual Threads (Project Loom): The built-in HTTP server uses lightweight Virtual Threads to handle concurrent search and ingest requests without blocking OS threads.
• Sharded Indexing: The indexer automatically shards data based on CPU cores for maximum lock-free write throughput.
• Built-in Text Analysis: Includes a custom Porter Stemmer and fast Murmur3 hashing for the term dictionary.