In some cases, you would be prompted to enter database credentials every time you published a workbook, even if you had previously provided database credentials in the same workbook session.
In some cases, excluding a mark more than once could change an associated filter from an include filter to an exclude filter.
In some cases, opening a workbook created in Tableau 9.1 in a newer version of Tableau Desktop would result in an internal error.
In the data source canvas, replacing an SAP HANA table with another table containing a special character in its name would result in an internal error.
Small decimals would sometimes appear as "null" in the data preview pane or show "#Error" when added to a view.
When connected to Microsoft Analysis Services, running a filter actions would sometimes cause Tableau Desktop to become unresponsive.
In some scenarios, involving cross-database joins and level of detail calculations, an "Unable to materialize temporary table" error would occur when you added a field to the view.
Specifying a negative value for number of decimals in the ROUND() function would cause Tableau Desktop and Tableau Server VizQL processes to error and crash.
Tableau Desktop would sometimes crash on Macs running macOS Sierra with integrated Intel graphics.
In some cases, when you edited a calculated field that was being used on Color, marks would lose their color and turn white or black.
It was not possible to union comma-delimited text files in locales where commas are used as decimal separators (e.g. German).
When connected to a SAS statistical file, IS8601DA10 formatted date fields were not correctly recognized and could not be converted to a different data type.
When you created a cross-database join involving a table with a back slash in its name, a Tableau Data Engine error would occur.
In some cases, certain calculated fields materialized in an extract prior to Tableau 10.0 could return more domain values than actually existed in the domain of the field when queried in Tableau 10.0 or newer.
Running filter actions could sometimes be significantly slower when compared to prior versions.
In some cases, deselecting filter actions did not behave as expected.
When connected to Google BigQuery, converting a string, containing a decimal character, to FLOAT would result in a "Bad int64 value" error.
When using the Refresh All Extracts functionality in Tableau Desktop, you would be prompted to enter a username and password for data sources that did not require credentials. Refreshes would still proceed and succeed if you canceled the unexpected prompts.
Updated Google BigQuery connections to utilize the REGEXP_CONTAINS function when REGEXP_MATCH is called.
Copying data from Tableau Desktop or exporting data from Tableau Server could multiply some decimal values by a factor of ten billion in the resulting data.