Agentic code converter went GA for SQL while SDKs added real-time warehouses and statement timeouts

October's Databricks SQL changes were mostly visible in SDK releases. A community post reported that the agentic code converter became generally available for SQL [1]. Most of the other product changes surfaced through API changes rather than announcements.

On October 8, three SDK releases (v0.187.0, v0.163.0 and v0.149.0) added statement timeout configuration for SQL warehouses [14][15][16]. Support for real-time SQL warehouses appeared in v0.167.0 on October 15 and in the Go SDK v0.189.0 on October 19 [10][8]. On October 21, releases v0.169.0, v0.190.0 and v0.155.0 added `customerVersionLabel`, `maxDbu` and `reydenClusterSizeMultiplier` to SQL warehouse create, edit and get responses [5][6][7]. Those same releases removed `replicationConfig` from `CreateShare` and `ShareInfo`, a breaking change for Sharing API users [5][6][7]. Earlier in the month, the SDK also gained custom template formats for SQL alerts [14].

Cost control was a recurring theme. A news article described tagging warehouses at creation and setting layered budgets with pacing alerts, using system.billing.usage to track unattributed spend [17]. A community thread asked how to set budgets and alerts on SQL warehouses for self-service users [12].

Practitioners also asked about behavior under load and about debugging. One thread asked what happens when many people query the same warehouse [11], and another asked how to debug remote_query() performance [2]. A separate write-up described rebuilding an 800-line PySpark pipeline in 200 lines of Databricks SQL [19], and a video covered Excel add-in functions such as databricks.sql [13].