When connected to Teradata, and using continuous date filters, Tableau Desktop would issue queries that cast date to timestamp resulting in poor query performance.
In some cases, changes made to dashboard layout in Tableau Desktop were not preserved when the workbook was saved and re-opened.
Adjusting a filter slider in Tableau Desktop for Mac would sometimes cause the application to crash.
In certain cases, when connected to an Excel data source containing pivoted fields and/or custom number formats, refreshing the live connection or extract resulted in seemingly random data being displayed in Tableau Desktop.
Labels for the Japanese cities of Minoh (箕面) and Ibaraki (茨木) were incorrect when the 'Place Names' map layer was displayed in Tableau Desktop.
In certain cases involving calculations with aggregate functions (e.g. COUNTD, MIN, MAX), Tableau Desktop query performance would differ dramatically when using a published data source vs. an identical local data source.
Wildcard filtering of string values was sometimes unacceptably slow.
When publishing a data source to Tableau Online from Tableau Desktop, the publish progress dialog window did not accurately show overall progress.
When pivoting multiple columns of different data types in Tableau Desktop, an error would occur: "Pivoted data type not unique for group".
Tableau Desktop would sometimes crash when opening a workbook with a web data connector data source.
When connected to a Hive data source in Tableau Desktop, queries containing the 'GROUP BY 1' clause would sometimes be extremely slow.
In some cases, changing a date filter selection would cause Tableau Desktop to crash.
In Tableau Desktop, using a set-based Top N filter with a separate context filter often returned incorrect results when the filters were modified.
In some cases, map shapes and other marks would render without any fill color.
Updated Australian postcodes.
Updated locations for the 33036 and 33050 zip codes.
Updated location for the 99603 zip code.
Tableau Desktop would intermittently crash when reading metadata.
Publishing a workbook containing visualizations with many mark labels could potentially be very slow.