1.12.1

dbt-databricks 1.12.1 (June 10, 2026)

Features

• Expose job_id, job_run_id, and task_run_id from the Databricks Jobs dbt_task runtime in adapter_response, enabling correlation between dbt runs and Databricks workflow executions via run_results.json by @sd-db in #1451, closes #722
• Add support for SPOG (Single Point of Gateway) hosts: account-level vanity URLs with ?o=<workspace-id> in http_path route correctly for both data-plane (SQL) and control-plane (REST/Jobs/Workspace API) traffic. Requires databricks-sql-connector >= 4.2.6 and databricks-sdk >= 0.76.0. by @sd-db in #1479

Fixes

• Ignore the server-set delta.parquet.compression.codec tblproperty when diffing relation configs, so streaming tables and materialized views are not flagged as changed after the backend started stamping it automatically by @sd-db in #1489
• Fix missing f-string prefix in JobRunsApi.submit debug log by @Dev-X25874 in #1471
• Fix capability-branching macros falling through to their legacy path at parse/compile time on SQL warehouses. The parse-time stub of has_dbr_capability now returns True on warehouse profiles for capabilities flagged sql_warehouse_supported, so macros select the modern branch during compilation instead of the legacy fallback. by @sd-db in #1449, closes #1331
• Fix snapshots not applying databricks_tags on columns by @themattmorris in #1442, closes #1441
• Skip DESCRIBE TABLE EXTENDED ... AS JSON for foreign/federated tables in get_columns_in_relation, avoiding repeated failures and extra latency on those sources by @willweld in #1472
• EXTRACT_CLUSTER_ID_FROM_HTTP_PATH_REGEX now stops the capture at ? / &, so any trailing query string on http_path no longer corrupts the extracted cluster id. Latent issue on legacy hosts; the fix unblocks SPOG cluster paths. by @sd-db in #1479
• Gate column-level constraints on contract.enforced to match the existing model-level gate, ensuring column-level NOT NULL / PK / FK / CHECK constraints are only applied when contract.enforced: true under use_materialization_v2: true by @sd-db in #1470, closes #1381
• Fix managed Iceberg incremental models configured with partition_by silently losing their clustering after the first incremental run. Managed Iceberg stores partition_by as liquid clustering server-side, so the reconciler now treats partition_by as the desired clustering and no longer issues a spurious ALTER TABLE ... CLUSTER BY NONE by @sd-db in #1496, closes #1495

Under the Hood

• Defer SDK Config construction to connection-open time so offline paths (dbt parse/list/compile) don't trigger the host-metadata probe introduced in databricks-sdk>=0.103; as a side effect, auth errors now surface at first connection rather than during profile parsing. by @sd-db in #1474
• Bump ceilings on databricks-sdk (now <0.105.0) and databricks-sql-connector[pyarrow] (now <4.3.0) to admit newer releases; floors unchanged. by @sd-db in #1474
• BREAKING: users who relied on column-level constraints (NOT NULL, primary key, foreign key, check) being applied under use_materialization_v2: true without contract.enforced: true must now set contract.enforced: true explicitly on the model. by @sd-db in #1470
• Bump upper bound of dbt-core to <1.11.12 to include dbt-core 1.11.9, 1.11.10, and 1.11.11 by @sd-db in #1505

Full Changelog: v1.12.0...v1.12.1