What databendlabs/databend shipped
Written by FoxPlug from public releases; not affiliated with Databend. An automatic summary of the public release, pull request and commit data of github.com/databendlabs/databend. Databend did not write it and does not use or endorse FoxPlug. Every line links to the public change it describes.
Get a weekly update like this for your product, free
Week of September 21, 2026
What shipped
- Release v1.2.949-nightly with CURRENT_TENANT_ID() context function, distributed query leak timeout configuration, and lineage capture for materialized views [0]. Release
- Tuple IN now supports multi-column subqueries with row-comparison semantics for comparing multiple columns against subquery results [1]. Pull request #20570
- Float zero and NaN values are now canonicalized in hashes and equality keys to prevent mismatches in Bloom indexes, DISTINCT, and hash joins [8]. Pull request #20547
- Webhook notifications now support WEBHOOK_BODY_TEMPLATE to wrap messages in caller-defined JSON body for Feishu, Slack, and DingTalk style bots [9]. Pull request #20539
- Row-level TTL DDL and metadata support added for writable persistent FUSE tables [29]. Pull request #20496
- Table locks are now released before ANALYZE and statistics regeneration to prevent blocking concurrent maintenance operations on large tables [3]. Pull request #20551
- Delta Lake table engine and its dependencies have been removed [18]. Pull request #20558
- Experimental virtual column support is now enabled by default for tables where virtual column option is set [2]. Pull request #20566
- Meta gRPC command-line clients now support TLS with --grpc-tls-ca-cert and --grpc-tls-domain-name flags [23]. Pull request #20555
- Three-valued logic fixes in equivalence folding and LIKE operations on constant operands prevent silent wrong results [13]. Pull request #20545
Why it matters
This week delivers critical correctness fixes for float semantics, query correlation, and three-valued logic that prevent silent wrong results. Feature additions include row-level TTL, webhook templating, and tuple IN with multi-column subqueries. Infrastructure improvements like virtual column defaults and TLS support enhance usability and security.
Changelog entry
- feat(query): add CURRENT_TENANT_ID() context function [0] Release
- fix(query): make distributed query leak timeout configurable [0] Release
- feat(query): capture lineage for materialized views [0] Release
- feat(query): support tuple IN with multi-column subqueries [1] Pull request #20570
- chore(query):
enable_experimental_virtual_columndefault value is now true [2] Pull request #20566 - fix(query): canonicalize float zero and NaN in hashes and equality keys [8] Pull request #20547
- feat(query): support WEBHOOK_BODY_TEMPLATE for notification [9] Pull request #20539
- feat(query): add row-level TTL DDL and metadata support [29] Pull request #20496
- fix(query): release table locks before analyze and rebase statistics [3] Pull request #20551
- fix(query): match nullable scalar correlation keys [4] Pull request #20408
- fix(query): three-valued logic in equivalence folding and LIKE on constant operands [13] Pull request #20545
- feat(meta): support gRPC TLS in command-line clients [23] Pull request #20555
- refactor(query): remove delta lake table engine and its dependencies [18] Pull request #20558
v1.2.949-nightly ships tuple IN multi-column subqueries, float canonicalization in hashes, row-level TTL, and webhook body templates. Float zero and NaN fixes prevent silent mismatches in joins and aggregates.
Databend v1.2.949-nightly includes critical correctness fixes and new capabilities. Float zero and NaN values are now properly canonicalized in hash operations, preventing silent mismatches in Bloom indexes and hash joins. Tuple IN now supports multi-column subqueries with SQL row-comparison semantics. Row-level TTL provides automatic data lifecycle management for FUSE tables. Webhook notifications gain templating support for integration with Feishu, Slack, and DingTalk. Meta clients now support gRPC TLS endpoints. Infrastructure improvements include virtual column defaults and fixes to correlation key matching.
Week of September 14, 2026
What shipped
- v1.2.947-nightly released with support for indexes on materialized views, inverted index bundles referencing sibling objects, and inverted index writing during block construction. Release
- Inverted indexes now stream large components out of memory during block construction instead of assembling in memory. Pull request #20538
- Inverted index segments can be merged without re-tokenizing, enabling recluster to carry inverted indexes efficiently. Pull request #20546
- Materialized views now support inverted, NGRAM, vector, and spatial indexes with synchronous index creation. Pull request #20361
- Fixed interval aggregate keys and nested column statistics through type-aware dispatch paths. Pull request #20544
- Window RANGE INTERVAL frames now correctly use the session time zone instead of always using UTC. Pull request #20533
- Dynamic tables are now supported as an independent Fuse-backed engine for multi-table derived results with explicit REFRESH. Pull request #20445
- Sleep function moved to async pipeline using Tokio timers, removing the three-second duration limit. Pull request #20493
- Block metadata is reshuffled after pruning to improve read parallelism for distributed Fuse scans. Pull request #20400
- Inverted index storage refactored to use Tantivy with quickwit feature for better performance and scalability. Pull request #20513
Why it matters
The week focused heavily on inverted index improvements across storage and query layers, enabling indexes on materialized views and more efficient index management during writes and merges. Several query execution fixes ensure correctness in window functions, aggregate operations, and materialized view rewrites. New capabilities like dynamic tables and async sleep expand functionality while storage optimizations like block metadata reshuffling improve distributed query performance.
Changelog entry
- Support indexes on materialized views (inverted, NGRAM, vector, spatial) Release
- Stream inverted index components during block construction instead of buffering in memory Pull request #20538
- Merge inverted index segments without re-tokenizing Pull request #20546
- Fix interval aggregate keys and nested column statistics in type dispatch Pull request #20544
- Fix window RANGE INTERVAL frames to use session time zone Pull request #20533
- Add dynamic table support with explicit REFRESH on Fuse engine Pull request #20445
- Make sleep asynchronous and remove duration limit Pull request #20493
- Reshuffle block metadata after pruning to improve read parallelism Pull request #20400
- Refactor inverted index storage to use Tantivy with quickwit Pull request #20513
- Expose aggregate function documentation in system.functions Pull request #20522
v1.2.947-nightly ships indexes on materialized views, streaming inverted index construction, and efficient index merging without re-tokenizing. Window functions now respect session timezones.
v1.2.947-nightly brings substantial improvements to indexing infrastructure. Inverted indexes now stream construction out-of-memory, merge without re-tokenizing, and work on materialized views. Query fixes ensure correct behavior in window RANGE INTERVAL frames and nested column statistics. Storage optimizations reshuffle block metadata post-pruning for better distributed read parallelism. Dynamic tables now supported as Fuse-backed multi-table derived results.