v0.5.2

Implements

• Add a write_operation hook to StorageCatalog so custom catalogs can run code before and after every write.

Fixes

• Only update docs page on release.

v0.5.1

Implements

• Renames repo to dbt-polars
• Unit tests defined in YAML, on SQL and Python models.
• The Copier template for custom catalogs moved into the dbt-polars repository: uvx copier copy gh:datamindedbe/dbt-polars <dir>.
• Full rewrite of the documentation for consistency and clarity.
• dbt.position() macro adapted to Polars SQL.
• dbt_expectations regex tests, through a dispatch setting in dbt_project.yml.

Fixes

• Incremental merge models on Delta tables no longer drop new columns with on_schema_change set to append_new_columns or sync_all_columns.
• Unsupported macros such as hash() now fail at execution time instead of at compilation time.
• dbt-polars did not recognize CTEs when they started with a comment.
• Change Polars dtype names to use double quotes instead of single quotes to generate valid SQL.
• Type checks (is_float, is_string, ...) did not recognize Polars type like Float64.
• Iceberg table models failed on Databricks when a column changed type, e.g. from integer to string.

v0.5.0

Implements

• [Databricks Catalog]: Add support for iceberg tables in the Databricks catalog.
• [Fabric Catalog]: New catalog for Delta and Iceberg tables in a Microsoft Fabric Lakehouse.
• [Snowflake Catalog]: New catalog for Snowflake-managed Iceberg tables through Snowflake Horizon Catalog.
• compute_threads in the profile limits how many of the threads models run Polars at the same time; models waiting for an Iceberg commit let others compute.
• Enable lazy evaluation of incremental models with sync_all_columns for delta and iceberg tables.

Fixes

• [Iceberg Catalog] Improve efficiency of merge and delete+insert models using composite keys (#45).
• [Iceberg Catalog] Merge and delete+insert no longer fail when a unique_key column contains NULL values.
• [Databricks Catalog] Dropping a Delta table on Azure now deletes its storage folder. Requires installing dbt-polars-catalog-databricks[azure].

v0.4.0

Implements

• A source can be read from a path relative to its catalog with external.location: a file with any extension (data.CSV) or a glob (orders/*.CSV) (#29)

Fixes

• Fail with error on SQL statements executed through hooks (#36)
• Resolve polar types in read options (schema, override_schema) (#33)
• Fail with error on unknown read_options keys (#34)
• Lazy append on incremental materialization (#31)
• Full refresh of an incremental model are now atomic
• Fail with error on unknown model config keys

v0.3.1

Implements

• current_timestamp macro evaluates to naive timestamp when dbt run started.
• dateadd, datediff, date_spine and get_intervals_between macro.
• date_trunc macro supports week and quarter.
• last_day macro for week, month, quarter and year.
• Move documentation to GitHub pages

Fixes

• Omit sensitive fields in the profile from dbt debug
• [Databricks Catalog]: Request READ instead of READ_WRITE credentials for read-only access
• [Databricks Catalog]: Update the description of all columns in a table with a single query
• Queried fail when a CTE has the same name as a model it selects from
• User .str.date to cast strings to date as this is the only supported way to do this conversion in Polars 2.0
• Throw an error when a user uses a hook or contract which is not yet supported

v0.3.0

Implements

• Catalogs are now published as separate packages
• Move tests into a package module (dbt.adapters.polars.testing) for reusability in catalogs maintained outside this repo
• [Databricks Catalog]: Add field table_format in the profile. Only allowed value is delta.

Other improvements

• Stop using pull_request_target in CI

Documentation

• [Readme] Add list of key dbt-polars features