Proposed Feature: Add automatic fallback from OData enum types to string in Fabric Copy Activity
Problem Description
When using Microsoft Fabric Data Factory’s Copy Activity with an OData v4 source, any column defined as an enum type in the OData service is silently dropped during schema inference and data ingestion. This occurs because Fabric Copy Activity performs strict metadata‑based type validation and only materializes properties that map directly to supported primitive types. Since OData enums are logical types rather than primitive scalars, they are excluded from the inferred schema.
This behavior results in incomplete datasets, unexpected data loss, and inconsistent ingestion outcomes—especially for systems that rely heavily on OData metadata and enum‑based domain modeling.
Expected Behavior
Power BI, which uses the same OData v4 metadata document for schema discovery, successfully handles enum types by gracefully coercing them into strings when no native type mapping exists. This demonstrates that a metadata‑driven fallback is technically feasible and consistent with Microsoft’s broader OData tooling ecosystem.
The expected behavior for Fabric Copy Activity is to follow the same principle:
If an OData enum type cannot be mapped to a native Fabric type,
Automatically fall back to string,
Instead of dropping the column entirely.
Why This Matters
Many enterprise OData services rely on enums to model business concepts such as status, category, priority, workflow state, and other domain‑specific classifications. Losing these fields during ingestion creates several issues:
Data loss: Critical columns disappear without warning.
Broken lineage: Downstream transformations and semantic models cannot rely on stable schemas.
Inconsistent behavior: Power BI and Fabric interpret the same OData metadata differently, causing confusion for users who expect parity.
Workarounds are impractical: Switching to Web API is not viable because it lacks metadata, type information, and relational structure required for OData‑based ingestion.
Requested Enhancement
Implement a fallback mechanism in Fabric Copy Activity so that:
OData enum types are automatically materialized as string columns.
Schema inference remains metadata‑driven and predictable.
No columns are silently dropped due to unsupported logical types.
Benefits to Microsoft Fabric Users
Consistent behavior across Microsoft products (Power BI and Fabric).
Improved reliability of OData ingestion pipelines.
No unexpected schema loss, reducing operational risk.
Better support for enterprise OData services, which frequently use enums.
Reduced need for custom workarounds, manual schema overrides, or alternative APIs.
Thank you for reading 😉
Recent ideas
Data Pipelines - Run only selected activities
For debugging and testing pipeline activities during development, allow us to select one or multiple activities and run only the selected pipeline activities. For example, I'm working on editing ...frithjof_v1 day agoCommunity ChampionNew619Views11likes2CommentsSemantic model connection bindings should be in source control (Git)
Semantic model data source connection bindings should be source controlled. A semantic model can contain multiple data source references, each of which can be mapped to a separate Fabric data connec...frithjof_v1 day agoCommunity ChampionNew21Views1like0CommentsBulk changing column names in Visualizations Pane
We often use raw/api column names or measures with a set nomenclature to be consistent and to keep track of them but we do not want to display these names in the visuals. Currently we have to change ...vishal1401971 day agoFrequent VisitorNew6Views0likes0Comments