Similar to Excel - RadPivotGrid could have an area, where you can select some filter/group descriptions.
If a group has only one item in it, I only want to show the total row and not being able to expand it.
As a user I want to hide the Grand Totals for specific groups.
In the RadPivotGrid: Open the Label Filter dialog on a dimension, don't set anything and press OK, then all items are present and nothing is filtered out as expected. Open the Label filter again and manually specify that the value must equal "xyz" and click OK then it will exclude every item that doesn't match this. Again, this behaviour is what you'd expect. If you then open the dialog, and delete the "xyz" from the textbox so that the dialog appears exactly as it did originally, and click OK then you won't see any items in your pivot grid. This is because it is now filtering out anything that doesn't have an empty string as the label. This is not what I'd expect. To remedy this, you can click on the dimension and see that the Label filter is still ticked in the context menu, and click "Clear filter", but this isn't exactly intuitive. I'd like the filter to be cleared when you clear the text in the box.
As a user I would like QueryableDataProvider to avoid blocking the UI while executing the query on the database server.
Currently when a header value is empty the displayed text is "(blank)" regardless of the application's culture. Localize this value or allow the developer to set this value more easily. If you want to manually localize the "(blank)" text you can define a RowHeaderTemplate and ColumnHeaderTemplate (or template selector) and replace the default text with its translation. You can do that via IValueConverter. (http://docs.telerik.com/devtools/wpf/controls/radpivotgrid/styles-and-templates/templating-cells#using-custom-cell-and-header-templates)
If you apply a number format to a Count value in the RadPivotGrid, then it is ignored. The reason given is that count is always an integer, so it shouldn't need formatting. However, I'd like to be able to apply group separators. It's much easier to read 100,000,000 than 100000000.
RadPivotGrid is loading all data from OLAP Cube. It should be able to use on demand (lazy loading) pattern, so when a group is expanded, a new query is sent and only required data is taken from the server.
Implement: "% of Parent", "% of Parent Row", "% of Parent Column"
I want to create CustomDistinct aggregate, I did it successfully for LocalDataSourceProvider, and I need it also for QueryableDataProvider. See: http://www.telerik.com/forums/custom-aggregate-for-queryabledataprovider