0.1.4 Release Notes

Release date: 2026-05-14

Summary

This release brings qql-go in line with the upstream QQL 2.2.0 feature set for point retrieval, pagination, turbo quantization, configurable hybrid fusion, and sparse rerank, while tightening release docs and validation behavior.

Highlights

SELECT support — SELECT * FROM <collection> WHERE id = <point_id> retrieves a single point by ID.
SCROLL support — SCROLL FROM <collection> [WHERE ...] [AFTER <id>] LIMIT <n> pages through points with optional filters and cursor offsets.
Hybrid DBSF fusion — SEARCH ... USING HYBRID FUSION 'dbsf' is now supported alongside the default RRF fusion mode.
Turbo quantization — CREATE COLLECTION ... QUANTIZE TURBO [BITS <1|1.5|2|4>] [ALWAYS RAM] is supported through the upgraded Qdrant Go SDK.
Sparse-only rerank — SEARCH ... USING SPARSE RERANK works on the cloud inference path.
Configurable dump batch size — qql-go dump --batch-size <n> controls how many points are emitted per INSERT BULK statement in dump output.

Supported syntax added in this release

New Syntax in 0.1.4

SELECT * FROM <name> WHERE id = '<uuid>' SELECT * FROM <name> WHERE id = <integer>
SCROLL FROM <name> LIMIT <n> SCROLL FROM <name> WHERE <filter> LIMIT <n> SCROLL FROM <name> AFTER '<point_id>' LIMIT <n> SCROLL FROM <name> WHERE <filter> AFTER <point_id> LIMIT <n>
SEARCH <name> SIMILAR TO '<query>' LIMIT <n> USING HYBRID FUSION 'dbsf' SEARCH <name> SIMILAR TO '<query>' LIMIT <n> USING SPARSE RERANK
CREATE COLLECTION <name> QUANTIZE TURBO CREATE COLLECTION <name> QUANTIZE TURBO BITS <1|1.5|2|4> CREATE COLLECTION <name> QUANTIZE TURBO BITS <1|1.5|2|4> ALWAYS RAM

Notes

Full changelog

See CHANGELOG.md.

For full commit list and details see the original release notes.