qdrant_edge_flutter 0.1.0

On-device vector search for Flutter, powered by the qdrant-edge Rust crate with built-in BM25 text embedding.

Changelog

0.1.0

• Initial release: an on-device Qdrant client for Flutter via dart:ffi over the qdrant-edge Rust engine.
• QdrantEdge client: createShard / loadShard, createBm25, createDense, openTextIndex, snapshot helpers.
• Shard with the full surface: upsert, deletePoints, search, query (prefetch + RRF/DBSF fusion), retrieve, scroll, count, info, facet, payload ops, field indexes, runtime named vectors, HNSW/optimizer config, flush, optimize, close. Complex arguments are plain Dart Map/List passed to the engine as JSON.
• On-device embedders: Bm25 (sparse, no model to ship) and Dense (MiniLM via candle, optional). TextIndex convenience for the add-text/search-text flow (lexical, or hybrid with a model).
• Prebuilt native binaries ship in the package (iOS dynamic QdrantEdgeFFI.xcframework, Android .so for arm64-v8a + x86_64, macOS .a) — no Rust toolchain required to consume the plugin.
• Platforms: Android, iOS, macOS.