Filters now go on a column-by-column basis. By going on a more row/tuple basis the logic can go to the next row as soon one of the filter parts gives back a False.
This should improve performance and stop the (for users strange) effect that adding filters makes the performance slower instead of faster.
Indexing should improve this further in the future, but for now this should improve things
Filters now go on a column-by-column basis. By going on a more row/tuple basis the logic can go to the next row as soon one of the filter parts gives back a False. This should improve performance and stop the (for users strange) effect that adding filters makes the performance slower instead of faster. Indexing should improve this further in the future, but for now this should improve things