v1.7.38
Version 1.7.38 includes bug fixes for ORDER BY aliases, Snowflake TIMESTAMP support, and React stale response handling, plus dbt manifest uploads and query orchestrator performance improvements for faster SQL tuple reuse.
1.7.38 (2026-09-14)
Bug Fixes
- cubesql: preserve literal ORDER BY aliases (#11496) (5019312)
- cubestore: Build with large pagesize support for AMR64, thanks @vltmn (#10152) (460173d)
- deps: Bump/align 24 transitive deps to clear 97 Dependabot alerts (#11846) (eb11ecf)
- react: ignore stale useCubeQuery responses (#11848) (ae50256)
- schema-compiler: name the ClickHouse types a cast can produce (#11807) (c7dcc5c)
- server-core: build one orchestrator api per id, not one per concurrent caller (#11834) (97282f0), closes #11661
- snowflake-driver: Support TIMESTAMP_TZ/TIMESTAMP_LTZ, thanks @npwalker (#11404) (81d2d23)
Features
Performance Improvements
Source: original entry ↗
More from Cube
Follow Cube to get its new changes in your feed and email digest.
Version bump for cubejs package
Version bump for the cubejs package. No functional changes included in this release.
Version 1.8.7
This release includes bug fixes for CLI placeholder versions, SQL segment filtering, timestamp arithmetic, and Trino driver timezone handling. New features add security context support to cube evals, COUNT(DISTINCT) planning over pre-aggregations, SQL push down to Cube Store, improved compiler caching, and join aliases. Query dialect resolution is now optimized per data source.
Stop rollupJoin and rollupLambda from leaking memory through memoized references
Fixed a memory leak in the schema-compiler where rollupJoin and rollupLambda functions were holding memoized references. This resolves performance degradation in long-running processes.