v1.11.2

What's Changed

• fix: pin click dependency version by @ericj-db in #1265
  • This was needed to fix a dependency resolution issue in Lakeflow, for which we yanked 1.11.1

v1.11.1

What's Changed

Features

• Add support for hard_deletes='new_record' in snapshot materializations, enabling tracking of deleted source records with dedicated deletion records marked by dbt_is_deleted column (thanks @randypitcherii!) (#1176, #1263)
  • Implements complete support for all three hard_deletes modes: ignore (default), invalidate, and new_record
  • new_record mode creates deletion records with actual source column values and dbt_is_deleted=true for full audit trail
  • invalidate mode uses Delta Lake's WHEN NOT MATCHED BY SOURCE clause to set dbt_valid_to on deleted records
  • Uses Databricks native BOOLEAN type for dbt_is_deleted column for improved type safety and performance

Fixes

• Fix bug that was applying UniForm tblproperties on managed Iceberg tables, causing materializations to fail
• Switch to a more reliable mechanism for checking schema existence (forward-ported from 1.10.15) (1261)

Under the hood

• Add validation for query tag value length and auto-escape special characters
• Add @@ prefix to system query tag keys (to keep them separate from user provided tags)

Full Changelog: v1.11.0...v1.11.1

v1.10.15

Fixes

• Switch to a more reliable mechanism for checking schema existence (1261)

Under the hood

• Allow for dbt-core 1.10.15 (1254)

v1.9.8

What's Changed

Fixes

• Switch to using full_data_type column when using info schema for column info (950)

Under the hood

• Pin dbt-core to 1.10.13 to prevent 1.9.X getting unexpected updates (1255)

Full Changelog: v1.9.7...v1.9.8

v1.11.0

Features

• Support databricks_tags for MV/STs
• Add support for scalar SQL functions (SQL UDFs) (1197)
• Add liquid clustering config for materialized views and streaming tables (thanks @reflection!) (1101)
• BEHAVIOR CHANGE: Add official support for insert_overwrite incremental strategy for SQL warehouses. This strategy now uses REPLACE ON syntax for all compute types (previously INSERT OVERWRITE). This behavior is gated behind behavior flag use_replace_on_for_insert_overwrite which defaults true (1025)
• Add support for Databricks query tags; NOTE: This feature is still in private preview, so it is not generally usable from the adapter yet.
• Add support for managed iceberg when table_format is set to iceberg. This behavior is gated behind behavior flag use_managed_iceberg which defaults to false
• Support delete+insert incremental strategy (thanks @canbekley!) (1217); if you have a current DBR (17.1+), this uses REPLACE ON for efficient replacement. Otherwise uses a 'delete from' statement follow by an insert statement (i.e. it is not atomic).

Fixes

• BREAKING: Fix column order mismatch bug in incremental models by using INSERT BY NAME syntax (#1211)
  • When using on_schema_change: sync_all_columns, dbt previously used positional column matching in INSERT statements, causing values to be inserted into wrong columns when column order changed
  • Now uses Databricks INSERT BY NAME syntax to match columns by name instead of position, preventing data corruption
  • Breaking Change: Requires Databricks Runtime 12.2 LTS or higher
  • Users on older runtimes should pin to dbt-databricks 1.10.x
  • Affects all incremental strategies: append, insert_overwrite, replace_where, and merge (via table creation)
• Fix case-sensitivity issues with column name handling in persist_docs and config diff operations (#1215)
  • Fixed KeyError when column names in models had different casing than YAML schema definitions
  • Improved efficiency of column tags and comments change detection to use case-insensitive comparison
• Use backtick quoting for everything to avoid errors with special characters (1186)
• Ensure column compare always uses lower case names (since Databricks stores internally as lower case) (1190)
• Fix incompatible schema error during streaming table creation (1235)
• Reintroduce support for external table type so as not to break users (1240)

Under the Hood

• Materialized views now uses CREATE OR REPLACE where appropriate, instead of DROP + CREATE
• Refactor to use Databricks SDK for API calls (1185)
• Update dependency versions, and start using uv (1199)
• Upgrade ruff and mypy (1207)
• Allow create or replace semantics on full refresh in Mat V2 (1210)
• Add centralized DBR capability system for managing version-dependent features with per-compute caching (#1218)
• BREAKING: Removing the 'use_info_schema_for_columns' behavior flag, as we have a better mechanism for getting complex type information - DESCRIBE EXTENDED ... AS JSON. This is a breaking change because it requires a modern DBR (or SQL Warehouse) in order to function (1226). If this is breaking for you, you will either need to upgrade your cluster to 16.4+ or pin dbt-databricks to 1.10.x.
• Use atomic CREATE OR REPLACE instead of DROP + CREATE for managed Iceberg tables
• BREAKING: Drop support for python 3.9, adds 3.13 (1240)
• BREAKING: Flipping the default for USE_USER_FOLDER_FOR_PYTHON to true (1248)

v1.10.14

What's Changed

• chore: Update dbt dependencies version ranges by @zqureshi in #1227

Full Changelog: v1.10.13...v1.10.14

v1.10.13

What's Changed

Fixes

• Fix issue causing MV/STs to always trigger as having their config changed (1181)
• Fix pydantic v2 deprecation warning "Valid config keys have changed in V2" (thanks @Korijn!) (1194)
• Fix snapshots not applying databricks_tags config (thanks @jeremyyeo!) (1192)
• Fix to respect varchar and char when using describe extended as json (1220)

Under the hood

• Update dependency versions (thanks @zqureshi!) (1223)
• Adding AGENTS.md ([1183]#1183)

New Contributors

• @Korijn made their first contribution in #1194
• @jeremyyeo made their first contribution in #1192
• @zqureshi made their first contribution in #1223

Full Changelog: v1.10.12...v1.10.13

v1.10.12

What's Changed

• chore: Bump pyproject.toml dependencies by @emmyoop in #1178

New Contributors

• @emmyoop made their first contribution in #1178

Full Changelog: v1.10.11...v1.10.12

v1.10.11

What's Changed

• fix: Set correct cluster type on databricks connection by @ericj-db in #1158
• fix: Handle ANSI mode for pandas DataFrame conversion by @benc-db in #1157
• chore: Modify integration workflow to simplify running external PRs by @benc-db in #1162
• fix: add check in _get to validate model had gettable config by @colin-rogers-dbt in #1164
• fix: Remove external path on intermediate tables for python incremental models by @gsolasab in #1161
• fix: get_columns_in_relation branching for STs by @ericj-db in #1167
• fix: Improve perf on schema enumeration/validation by @benc-db in #1168
• fix: don't pass nodes that don't have database attribute to build_catalog relation by @colin-rogers-dbt in #1170
• Fix: model-level override logic for is_cluster by @ericj-db in #1172

Full Changelog: v1.10.10...v1.10.11

v1.10.10

What's Changed

• DOCS: Update Databricks Workflow to match new naming Databricks Lakeflow Jobs by @excavator-matt in #1145
• write tables to different catalogs by @colin-rogers-dbt in #1129
• feat(1111): add pr linting to enforce conventional commits by @Jeremynadal33 in #1126
• chore: Tighten version range for higher risk dependencies by @ericj-db in #1147
• fix: COMMENT ON syntax not supported prior to 16.1 by @benc-db in #1151

New Contributors

• @excavator-matt made their first contribution in #1145

Full Changelog: v1.10.9...v1.10.10