You are here:
Current Limitations for Semantic Models Created from a PDS in Tableau Cloud
Semantic models created from Published Data Source (PDS) in Tableau Cloud currently have the following limitations.
Query Capabilities
- Fiscal calendar isn’t supported. However, you can manually create a calculated field for using the DateAdd function to refresh and align data with the fiscal year.
- Filtering with the like operator isn’t supported.
- The Row Count feature isn’t available. See Row Count in Tableau Semantics for more information.
- Table calculations created in the PDS aren’t yet reflected in the Semantic Model. As a temporary solution, recreate the same calculation directly in the Tableau Semantics layer.
Tableau Next Flows
- Extending PDS semantic models isn’t available.
Semantic Model
- Creating groups or bins isn’t supported.
- Adding new Data 360 objects to the model isn’t available in the current version.
- Creating a Logical View (LV) isn’t supported.
- Groups and bins defined in the PDS aren’t temporarily inherited.
- DateTime-type parameters aren’t inherited from the PDS.
- Modifying field names, default aggregation, or types isn’t available.
Datetime/Time Zone Fields
When integrating PDS with Tableau Next, date and time values may appear inconsistent across visualizations due to a mismatch in time zone handling between Tableau Cloud and Tableau Next.
- Tableau Cloud (the source) uses “time zone naive” (Wall Clock) time, which treats the date and time exactly as they were entered.
- Tableau Next (the client) is “time zone aware”. It assumes all incoming data is in UTC and automatically applies a local time zone shift based on the user's browser.
The difference in logic causes a single data point to show two different values on the same page. For instance, a summary chart might correctly attribute a record to 2025 by using the literal source value, whereas a detailed list view might shift the same record backward due to the browser's local time zone offset. A specific example is a midnight entry for January 1, 2025, which would incorrectly appear as December 31, 2024, for a user in a New York (UTC-5) time zone.
As a temporary workaround, manually set your browser or user time zone to UTC. This prevents the system from automatically shifting dates and makes sure that all views consistently match the original "Wall Clock" time of the source data.
