dbt feature compatibility

Overview of features dbt-polars already supports and plans to support in the future.

Status	Meaning
✅	Supported
🟡	Partially supported
🚧	Not supported yet, but planned
❌️	Not supported, unplanned

dbt core features

Feature	Status	Remarks
table materialization	✅
incremental materialization	🟡	✅ append ✅ merge ✅ delete+insert 🚧 microbatch
ephemeral materialization	✅
view materialization	❌️	dbt-polars has no database to hold views. Use table or ephemeral instead.
Snapshots	✅
Seeds	✅
Sources	✅
Python models	✅
Data tests (generic and singular SQL)	✅
Python singular tests	✅
Unit tests (defined in YAML)	✅
persist_docs	✅
dbt docs generate	✅
dbt show	✅
dbt clone	❌️
Hooks	❌️
Grants	❌️
Model contracts	🚧

dbt core extensions

dbt-polars adds three features that are not supported in dbt-core:

• Ephemeral Python models
• Python data tests
• Unit tests on Python models

Macros

dbt-core publishes a list of cross-database macros that each warehouse should implement.

Macro group	Status	Remarks
Date/time	✅
Hash	❌️
Type conversion	✅
Strings	✅
Arrays	🟡	Missing: - array_append - array_concat
Aggregates	🟡	Missing: listagg with an order or limit
Set operations	✅
Other	✅

Unsupported macros raise a compilation error. Use a Python model for the same logic.

dbt packages

The table counts how many functions (macros and generic tests) of popular dbt packages work on dbt-polars.

Package	Tested version	Supported	Partially supported	Not supported
dbt_utils	1.4.1	39	0	5
codegen	0.14.1	4	0	1
audit_helper	0.11.0	5	1	1
dbt_expectations	0.10.10	58	0	6

dbt_utils

Not supported:

• generate_surrogate_key needs a hash function, which Polars SQL doesn't have.
• get_relations_by_pattern and get_relations_by_prefix query information_schema, which dbt-polars doesn't provide.
• at_least_one uses a with clause inside a subquery and having without group by. Polars SQL supports neither.
• not_constant uses having without group by. It works when you set group_by_columns.

codegen

Not supported:

• generate_source lists tables through information_schema. It works when you pass table_names.

audit_helper

Not supported:

• compare_all_columns uses a with clause inside a subquery, which Polars SQL doesn't support.

Partially supported:

• compare_relation_columns returns an empty column_name for columns that only exist in b_relation.

Primary keys made of several columns don't work in any macro, because combining them needs a hash function.

dbt_expectations

The regex tests (expect_column_values_to_match_regex, expect_column_values_to_not_match_regex and their _list variants) need a dbt-polars version of a dbt_expectations macro. Add this to your dbt_project.yml to use it:

dispatch:
  - macro_namespace: dbt_expectations
    search_order: ["dbt", "dbt_expectations"]

Not supported:

• expect_column_median_to_be_between and expect_column_quantile_values_to_be_between use percentile_cont, which Polars SQL doesn't have.
• expect_row_values_to_have_recent_data and expect_grouped_row_values_to_have_recent_data use dbt_date's current time, which needs convert_timezone.
• expect_column_values_to_be_within_n_moving_stdevs uses a rows between window frame, which Polars SQL doesn't support.
• expect_select_column_values_to_be_unique_within_record needs a hash function.