Data from a datasource or another view can be filtered and sorted before being passed to the view. Certain records can also be extracted, for example the top ten or bottom five can be retained and the rest discarded (this extraction operation occurs after any filtering and sorting).
Filtering allows you to define rules that each record is tested against. A typical rule might be: Country Equals "Singapore" or Salary Greater Than 2000. Only those records that pass all the rules are allowed to proceed. The filter criteria used here are the same as those used in the Filter Processor in the Data Designer manual. Please see that discussion for full details on the options available.
Sorting allows you to sort the records into ascending or descending order based on the values in fields. Multi-level sorting is supported, so you can sort by Country and then City and the tool will ensure that the Cities are sorted within their respective Countries. Contents such as Chart, Grid, Table and Cube have sorting capabilities now. Sorting within the view will override any earlier sorting, but will occur after any extraction. For example, you can sort sales people by performance, extract the top ten and then show them in a chart sorted in alphabetical order. Two variants of sort are supported, regular Ascending and Descending and Simple Ascending and Simple Descending. The Simple options are very fast, but don't take into account complex character differences, for example letters with accents produced by merged Unicode character combinations. The regular Ascending and Descending use the full Unicode algorithms. Further, the Simple variants treat whitespace as significant, whereas the Unicode algorithms don't (just like a dictionary).
Extraction is the retention of certain records based on their position in the data set. You might choose to extract those records at the top or the bottom of the set (presumably the order is significant, probably because they have just been sorted). You can choose to retain either a fixed number or a percentage of all the records. If a percent value is chosen, then the value will be "rounded up" to include partial records. For example, if you choose 1%, you will still get one record, even if there are only twenty records in the set.