Release 0.234¶

General Changes¶

Fix query failure for cross-joining bucketed tables with COALESCE when the bucket counts are compatible but mismatched.
Fix an issue where IGNORE NULLS is not respected for window functions.
Fix scheduling regression by setting the default scheduler to legacy scheduler.
Improve coordinator reliability by adding support to cap the number of total tasks running in a coordinator by pausing scheduling. This can be enabled by the configuration property experimental.max-total-running-task-count-to-not-execute-new-query.
Improve the scale writer heuristics by considering overall producer buffer utilization. This can be enabled by the session property optimized_scale_writer_producer_buffer and the configuration property optimized-scale-writer-producer-buffer.
Improve end point v1/resourceGroupState to supporting filtering of resource groups that are dynamically generated.
Improve connection pooling to avoid running out of sockets.
Add KHyperLogLog type and related functions.
Add support for forcing streaming exchange for Mark Distinct even if materialized exchange is enabled. This can be enabled by the session property use_stream_exchange_for_mark_distinct and the configuration property query.use-streaming-exchange-for-mark-distinct. (#14216).

Improve storage efficiency for Bing tiles by storing Bing tiles as BIGINT. This also reduces bucket skew in certain cases.
Add support for spatial joins for join condition ST_Distance(p1, p2) < r.

Add ZSTD support for writing ORC and DWRF files. This can be enabled by setting session property hive.compression_codec to ZSTD.

Add support for verifying SELECT queries that produce DATE or UNKNOWN (null) columns, or structured typed columns with DATE or UNKNOWN.
Add support for auto-resolving control check query failures due to EXCEEDED_TIME_LIMIT.
Add determinism analysis support for simple queries with top-level ORDER BY LIMIT clause. (#14181).