The node allows for row splitting according to certain criteria. It can include or exclude: certain ranges (by row number), rows with a certain row ID, and rows with a certain value in a the selected columns (attribute).
The RowFilter node has been adapted to allow for multiple column selection to perform the attribute matching (include and exclude). It is possible to match based on a single column match (only one must return true) or every column match (all must return true).
Example usage: String column with numbers, int column and a double column. Using the pattern 1 will return matches for "1" in the int column and the String column however will not match the double column. To match the value '1' on a double column the pattern needs to be '1.0'. If you want the same match to work accross these three columns then you can use the regex: '(1|1\.0)' (no quotes) which makes the '.0' optional.
Range filtering on mixed column types is not recommended. It is best to use all Int columns, all Double columns or all String columns etc. Although filtering on Strings will likely give unexpected results.