Skip to content

SQL Anywhere 0.5.0 — search as one chapter

Choose a tag to compare

View all tags
@kwhorne kwhorne released this 16 Jul 08:54
· 5 commits to main since this release
v0.5.0
54d255c
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

SQL Anywhere 0.5.0 — search, made whole and made honest

Highlights

  • Search is one chapter, not a separate product. Full-text (FTS5), faceted
    (GROUP BY), vector (DiskANN) and hybrid (RRF) search are the same engine,
    composed in plain SQL over the same rows — no dedicated search service to
    deploy or keep consistent. New guide: docs/SEARCH.md. Faceted search is
    demonstrated (examples/faceted_search.rs) and verified
    (tests/faceted_search.rs).
  • Real semantic embeddings, first-class. A pluggable Embedder trait lets
    you feed any model (local ONNX/candle or a hosted API) into the same
    vector32(...) + vector_top_k path. The dependency-free embed() is now the
    LexicalEmbedder. A worked neural example (examples/semantic-search, kept out
    of the core build) finds "the cat sat on the mat" for "a small feline rested
    on a rug"     — something lexical search cannot.
  • Hardened Docker releases. Every published image is smoke-tested on both
    amd64 and arm64 (scripts/smoke-test-docker.sh) — it boots the container
    and runs a real vector search before the release is considered good.

Install

docker run -p 8080:8080 -d ghcr.io/kwhorne/sqlanywhere-server:0.5.0

Prebuilt sqld binaries and crsqlite extensions for macOS Apple Silicon,
Ubuntu Intel and Ubuntu ARM are attached below.

Full changelog: https://github.com/kwhorne/sql-anywhere/blob/main/CHANGELOG.md

Assets 8
Loading