NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #5284 most downloaded on PyPI
A Microsoft Fabric Synapse Data Warehouse adapter plugin for dbt
Last release today
04 Oct 2026
Release timing varies
gaps range from 2 weeks to 5 months
Nearly every release is documented
notes for 32 of 33 stable releases
1 version withdrawn
withdrawn after publishing
3 years old
43 releases · first in 2023
One column per quarter.
Nothing published for this version
Enhance table refreshes, fix service principal auth, and alias rendering by pradeepsrikakolapu in #438
Full Changelog: v1.11.1...v1.11.2rc1
Warehouse transaction support — model, seed, snapshot, function, hook,
grant, constraint, statistics, and table-clone operations now participate in
dbt-managed BEGIN/COMMIT/ROLLBACK boundaries. Failed table reloads,
relation replacements, incremental writes, schema changes, seed resets, and
table or function creation restore the previous target instead of leaving
partial data or temporary relations. Clone and function grants,
documentation metadata, and transactional hooks commit with their resources.
Statements inside an open transaction are not retried individually, avoiding
replay of non-idempotent work.
Transaction-safe metadata queries — read-only relation, column, index,
freshness, and catalog queries no longer open transactions. This prevents
adapter introspection during operations such as
dbt_external_tables.stage_external_sources from accidentally owning and
rolling back subsequent DDL when the operation connection closes.
Schema-aware full refreshes — table models and incremental models running with
--full-refresh preserve the existing table object when its ordered schema, identity
properties, and cluster_by layout are unchanged. The adapter performs an atomic
TRUNCATE and full reload, retaining object-bound metadata and optimization history.
Named primary-key, unique, and foreign-key constraints are reconciled transactionally.
Schema, identity, physical-layout, or non-reconcilable custom-constraint changes use an
atomic CTAS/drop/rename replacement.
mssql-python — retain support for the
ServicePrincipal profile alias and stop adding the unsupported Authority Id
connection-string keyword. ActiveDirectoryServicePrincipal now uses the
Authentication, UID, and PWD fields supported by mssql-python. Resolves
#434.Add Python 3.11 support and update CI workflows by pradeepsrikakolapu in #436
Full Changelog: v1.11.0...v1.11.1
Upgrading v1.11.0 by pradeepsrikakolapu in #428
function(), support default arguments, and are discoverable through the adapter relation cache.dbt-core requirement from >=1.10.0 to >=1.11.0.dbt-adapters requirement from >=1.10.0,<2.0 to >=1.15.5,<2.0.1.11.latest branch.Port Fabric Warehouse changes from fabric-toolbox PR #552 by pradeepsrikakolapu in #407
Full Changelog: v1.10.0...v1.10.1
V1.10 update by pradeepsrikakolapu in #369
--sample flag) — supported by virtue of the existing microbatch incremental strategy implementation. Developers can run dbt run --sample="3 days" (or a static date range) to build time-bounded slices of models, reducing dev/CI runtimes and warehouse cost. Set event_time on sampled models to the timestamp field used for windowing. Resolves #313.dbt-core requirement from >=1.8.0 to >=1.10.0dbt-adapters requirement from >=1.1.1,<2.0 to >=1.10.0,<2.0v1.9.10 update by pradeepsrikakolapu in #368
Incremental merge: delete_not_matched_by_source — adds WHEN NOT MATCHED BY SOURCE THEN DELETE to the T-SQL MERGE statement. Deletes rows in the target whose unique_key is absent from the source relation. Use when the incremental model returns the complete current dataset (not a delta). Set delete_not_matched_by_source: true in the model config alongside incremental_strategy: merge. Resolves #361.
Incremental merge: delete_condition — issues a DELETE … FROM … INNER JOIN … WHERE statement after the MERGE, removing target rows that match the source on unique_key and satisfy a user-supplied SQL expression. Use for soft-delete patterns where the source carries a delete-flag column. Set delete_condition: "DBT_INTERNAL_SOURCE.is_deleted = 1" (or equivalent) in the model config alongside incremental_strategy: merge. Resolves #361.
fabric__split_part macro — replaced invalid CTE-inside-SELECT-clause with a derived-table subquery, and fixed a double-comma syntax error in the CTE SELECT list. split_part() now works correctly as a scalar column expression. Resolves #358.
Source freshness failures on Fabric Lakehouse — fabric__get_relation_last_modified now emits USE DATABASE before the query (was missing, causing cross-database resolution failures) and casts o.modify_date to datetime2(3) explicitly. A new fabric__collect_freshness macro overrides dbt's default using TRY_CAST(loaded_at_field AS datetime2(3)), safely handling Lakehouse sources where datetime columns are stored as VARCHAR (e.g. Debezium CDC timestamps). Addresses #364.
Suppress spurious ROLLBACK on connection close — overrides close() in FabricConnectionManager to set transaction_open = False before delegating to the parent. With autocommit=True there is nothing to roll back; the parent's ROLLBACK call was blocking for up to 11 minutes on Fabric Warehouse when concurrent DDL held catalog locks. Resolves #362.
Retry list_relations_without_caching on failure — overrides list_relations_without_caching in FabricAdapter with exponential back-off retry (1 s, 2 s, 4 s … capped at 30 s), using the existing retries credential. When query_timeout is set and a blocked catalog read times out, the adapter retries rather than failing the run immediately. Resolves #362.
Configurable connection pooling — adds a pooling credential (true by default, matching the previous hardcoded behaviour). Set pooling: false to disable pyodbc connection pooling, useful when routing through a proxy or when pool reuse contributes to catalog-lock contention. Resolves #362.
Performance guidance for large projects — added a README section recommending cache_selected_only: true, low thread counts (4–8), and query_timeout for projects with 500+ models or concurrent dbt runs on the same warehouse. Resolves #362.
Push schema filters into CTE source branches — moved the WHERE SCHEMA_NAME(...) like predicate from the outer select * from base into each individual sys.tables and sys.views branch in fabric__list_relations_without_caching and fabric__get_relation_without_caching. Filtering at the source narrows catalog scans and reduces exposure to row-level lock contention from concurrent DDL. Addresses #365.
macOS Apple Silicon install guidance — added a collapsible macOS (Apple Silicon) section to the README explaining the libodbc.2.dylib / libodbc.3.dylib version mismatch between pyodbc wheels and modern Homebrew unixODBC, with both the recompile fix and the symlink workaround. Resolves #351.
pyodbc import error message — wrapped import pyodbc in a try/except ImportError that re-raises with a clear, actionable error message (including the fix commands) before dbt's error handling can obscure it. Resolves #351.
Add openrowset, quoting/aliasing, and data clustering features by pradeepsrikakolapu in #355
Full Changelog: v1.9.8...v1.9.9
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.9.8...v1.9.9
Added merge capabilities and enabled warehouse snapshots by pradeepsrikakolapu in #328
Full Changelog: v1.9.6...v1.9.8
Nothing published for this version
V 1.9.6 update by Pradeep Srikakolapu (@prdpsvs) in #315
Full Changelog: v1.9.5...v1.9.6
V1.9.5 release by Pradeep Srikakolapu (@prdpsvs) in #295
Full Changelog: v1.9.4...v1.9.5
V1.9.4 by Pradeep Srikakolapu (@prdpsvs) in #283 , #284 - - reverting nested CTE changes
Full Changelog: v1.9.3...v1.9.4
split part fix for -ve part numbers and removed generate_custom_schema macro in the adapter by @prdpsvs in https://github.com/microsoft/dbt-fabric/pul
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.9.2...v1.9.3
V1.9.2 by @prdpsvs in https://github.com/microsoft/dbt-fabric/pull/264
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.9.1...v1.9.2
## What's Changed * #260 - Adding support of ephemeral models with Tables. Ephemeral models with NESTED CTE are not supported by views (not supported
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.9.0...v1.9.1
Added a new incremental strategy named "microbatch" in the valid_incremental_strategies method of fabric_adapter.py and corresponding SQL macros. [[1]
valid_incremental_strategies method of fabric_adapter.py and corresponding SQL macros. [1]fabric__build_snapshot_table and fabric__snapshot_staging_table macros to use configurable snapshot column names, enhancing flexibility and readability. [1] [2]fabric__snapshot_merge_sql macro to handle dynamic column names and added checks for dbt_valid_to_current. [1]snapshot.sql to integrate the new snapshot staging table logic and ensure proper column handling. [1] [2]test_incremental_microbatch.py, including setup for input models and insertion of test data.Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.8.9...v1.9.0
`.github/workflows/integration-tests-azure.yml`: Renamed the job and updated the profile and permissions for integration tests. Switched from Azure CL
.github/workflows/integration-tests-azure.yml: Renamed the job and updated the profile and permissions for integration tests. Switched from Azure CLI login to Azure login with OIDC and added a step to test the connection to Fabric Data Warehouse. [1] [2].github/workflows/publish-docker.yml: Added workflow_dispatch trigger and a step to list Docker images. [1] [2].github/workflows/unit-tests.yml: Limited Python versions to 3.10 and 3.11 for unit tests.dbt/adapters/fabric/fabric_connection_manager.py: Added methods for handling access tokens and updated the connection string logic to include pooling and retry configurations. [1] [2] [3] [4] [5] [6]dbt/adapters/fabric/__version__.py: Bumped version to 1.8.9.dbt/adapters/fabric/fabric_credentials.py: Added access_token attribute and updated default retry count.dbt/include/fabric/macros/adapters/catalog.sql: Added database usage checks and improved join conditions. [1] [2] [3] [4] [5]dbt/include/fabric/macros/adapters/metadata.sql: Ensured database names are properly formatted.dbt/include/fabric/macros/adapters/schema.sql: Added database usage commands before dropping schemas.dbt/include/fabric/macros/adapters/show.sql: Improved SQL limit handling logic.dbt/include/fabric/macros/materializations/models/table/table.sql: Enhanced logic for creating backup relations.dbt/include/fabric/macros/materializations/snapshots/helpers.sql: Simplified column creation logic.dbt/include/fabric/macros/materializations/snapshots/snapshot.sql: Improved handling of temporary views in snapshots.These changes collectively enhance the integration testing setup, streamline connection management, and improve the reliability and clarity of macro definitions.
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.8.8...v1.8.9
Even though this is a minor release, a lot of work has been done to secure the repository and code as part of Security Foundation Initiative by Micros
Even though this is a minor release, a lot of work has been done to secure the repository and code as part of Security Foundation Initiative by Microsoft in addition to addressing issues from the community. Thanks for your patience.
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.8.7...v1.8.8
Addressing issues #189, #188, #181, #179, #197 by @prdpsvs in https://github.com/microsoft/dbt-fabric/pull/192
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.8.6...v1.8.7
Updating dispatching methods to ensure dbt-synapse adapter can use fabric materializations
by @prdpsvs in https://github.com/microsoft/dbt-fabric/pull/178
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.8.5...v1.8.6
Unpinning adapter dependency to 1.8.0 by @prdpsvs in https://github.com/microsoft/dbt-fabric/pull/175
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.8.4...v1.8.5
V1.8.4 by @prdpsvs in https://github.com/microsoft/dbt-fabric/pull/172 Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.8.3...v1.8.4
Added Fabric Relation to suppport render_limited in 1.8 release, @prdpsvs in https://github.com/microsoft/dbt-fabric/pull/156
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.8.2...v1.8.3
Adding Synapse Spark Authentication Option by @marvinbuss in https://github.com/microsoft/dbt-fabric/pull/155
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.8.1...v1.8.2
Fixed snapshot related bug which is not working with CTE's by @prdpsvs in https://github.com/microsoft/dbt-fabric/pull/153
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.8.0...v1.8.1
V1.7.1 by @prdpsvs in https://github.com/microsoft/dbt-fabric/pull/100
Full Changelog: https://github.com/microsoft/dbt-fabric/compare/v1.7.4...v1.8.0
https://github.com/microsoft/dbt-fabric/compare/v1.8.0rc1...v1.8.0rc3
https://github.com/microsoft/dbt-fabric/compare/v1.8.0rc1...v1.8.0rc3
https://github.com/microsoft/dbt-fabric/compare/v1.7.latest...v1.8.0rc2
https://github.com/microsoft/dbt-fabric/compare/v1.7.latest...v1.8.0rc2
make devtests/functional/adapter/test_query_comment.py::TestMacroArgsQueryComments::test_matches_comment to use correct dbt_version, see dbt-coreSupporting dbt-core 1.8.0
Decouple imports to common dbt core and dbt adapter interface packages for future maintainability and extensibility.
From now on, Apple-silicon users don't have to locally build pyodbc, because M1, M2 binaries is included in pyodbc from 5.1.0 onwards!
use pre-release v1.8.0rc2 instead
Supporting dbt-core 1.8.0
Decouple imports to common dbt core and dbt adapter interface packages for future maintainability and extensibility.
From now on, Apple-silicon users don't have to locally build pyodbc, because M1, M2 binaries is included in pyodbc from 5.1.0 onwards!
Added Connect Retry Count to include in-built driver support
https://github.com/microsoft/dbt-fabric/compare/v1.7.2...v1.7.latest
Changelog
https://github.com/microsoft/dbt-fabric/compare/v1.7.2...v1.7.latest
https://github.com/microsoft/dbt-fabric/compare/v1.7.1...v1.7.latest
https://github.com/microsoft/dbt-fabric/compare/v1.7.1...v1.7.latest
https://github.com/microsoft/dbt-fabric/compare/v1.7.0...v1.7.latest
https://github.com/microsoft/dbt-fabric/compare/v1.7.0...v1.7.latest
Bump from pytest==7.4.2 to pytest==7.4.3
Allowing users to provide "ServicePrincipal" as an authentication mode. Adapter will treat ServicePrincipal as ActiveDirectoryServicePrincipal authent
https://github.com/microsoft/dbt-fabric/compare/v1.6.0...v1.6.latest
https://github.com/microsoft/dbt-fabric/compare/v1.6.0...v1.6.latest
Adding limit - new args to adapter.execute() function
Releasing 1.5.0 for dbt cloud integration
Upgraded dbt-fabric adapter to match dbt-core & dbt-tests-adapter version 1.5.2.
SP_RENAME is supported in FabricA minor release to follow up on the 1.4 releases.
A minor release to follow up on the 1.4 releases.
For full changelog: https://github.com/microsoft/dbt-fabric/blob/main/CHANGELOG.md
Updated connection property to track dbt telemetry by Microsoft.
A minor release to follow up on the 1.4 releases.
A minor release to follow up on the 1.4 releases.
For full changelog: https://github.com/microsoft/dbt-fabric/blob/main/CHANGELOG.md
Fixed view rename relation macro. Bumped required python packages versions.
Requires dbt 1.4.5 and previous versions are not supported by Fabric Data Warehouse. Microsoft is actively releasing/adding T-SQL support. Please rais
Requires dbt 1.4.5 and previous versions are not supported by Fabric Data Warehouse. Microsoft is actively releasing/adding T-SQL support. Please raise issues in case of any bugs.
DBT Supported features All materializations and resource features such as Tables, Views, Seeds, sources, tests and dbt docs are supported. Advanced features such as incremental and snapshot features may work but are not planned to support in 1.4.5. We recommend you read Microsoft Fabric Data Warehouse documentation before using the adapter.
Requires dbt 1.4.5 and previous versions are not supported by Fabric Data Warehouse. Microsoft is actively releasing/adding T-SQL support. Please raise issues in case of any bugs.
We recommend you to read Microsoft Fabric Data Warehouse documentation before using the adapter.
Your coding agent can read these notes before it upgrades. Set up the MCP server →