Forum Discussion
Dataverse Shortcut Delta Table Schema Not Updating After Source Column Precision Change
- 6 months ago
Hi akshaykhaneja ,
Dataverse shortcuts in Fabric infer and lock the Delta schema at creation time. Subsequent Dataverse metadata changes (such as increasing decimal precision) do not propagate to the shortcut schema. Refresh only reloads data using the existing schema, which is why values are truncated. Currently the only way to update the schema is to recreate the shortcut, or to cast the columns to higher precision in a downstream Silver layer.
If this post helps, then please appreciate giving a Kudos or accepting as a Solution to help the other members find it more quickly.
If I misunderstand your needs or you still have problems on it, please feel free to let us know. Thanks a lot! - 6 months ago
Hi akshaykhaneja,
Thank you ssrithar for your response to the query.
Alongside the previously mentioned points, it's important to highlight that this behavior is intentional in the current Fabric Dataverse shortcut implementation, and currently, in-place schema evolution for shortcuts is not supported. Additionally, there is no automatic detection or alerting if Dataverse metadata (like precision or scale changes) differs from the locked Delta schema, which means data accuracy can be affected if not manually checked.
For governance and operations, it’s best to treat Dataverse shortcuts as schema-static ingestion objects and add a validation step to your data pipeline whenever Dataverse column definitions are updated. If recreating the schema isn’t possible due to downstream dependencies, another approach is to store raw values as strings in the Bronze layer (when feasible) and apply correct typing at the Silver or Gold layers to prevent irreversible truncation.
Thank you.
Hi akshaykhaneja ,
Dataverse shortcuts in Fabric infer and lock the Delta schema at creation time. Subsequent Dataverse metadata changes (such as increasing decimal precision) do not propagate to the shortcut schema. Refresh only reloads data using the existing schema, which is why values are truncated. Currently the only way to update the schema is to recreate the shortcut, or to cast the columns to higher precision in a downstream Silver layer.
If this post helps, then please appreciate giving a Kudos or accepting as a Solution to help the other members find it more quickly.
If I misunderstand your needs or you still have problems on it, please feel free to let us know. Thanks a lot!