This is crucial functionality for dbt users working with Fabric. I'm hitting the same issue in my implementation.
The current setup is frustrating - I can move data between workspaces with pipelines but can't directly query across workspaces. This kills a clean implementation for data privacy requirements where:
- Raw data lives in workspace A with masking and viewer-only access
- Transformed data belongs in workspace B
- dbt needs to read from A but write to B
The workarounds all feel hacky:
- Shortcuts work but add complexity and management overhead
- Cross-warehouse queries only function within the same workspace
- Data Pipelines require completely different tooling/orchestration
- Notebook T-SQL queries are possible but that's another thing to maintain
Bottom line: If I'm a member of both workspaces, I should be able to read from one and write to another in the same dbt process. This would align with dbt's core philosophy of modeling data where it belongs while respecting security boundaries. Has anyone found a cleaner solution than shortcuts? Would love to see the dbt-fabric adapter support configuring separate source/target workspaces in profiles.yml.
