I observed an unexpected behaviour of “Color Manager”:
In my workflow I read data from database, filter via “Row Filter” for a special entry (in my case [NULL]), filtered data after Row Filter is ok (exclude these entries).
But in the following Color Manager node I still see the (excluded) filtered entries with “[NULL]”.
If I filter the entries direct in the also used “DB Query Reader” then everything it’s ok.
Can anyone explain why Color Manager seems to ignore filtered rows via Row Filter?
Hi,
The Row Filter does not update the domain of the data. To force that, you need to insert a Domain Calculator node after it.
Kind regards,
Alexander
Hi Alexander, thanks for this very quick response.
I’ve tested with Domain Calculator and yes that works.
But can you please explain? The row filter excludes excactly the rows I want to filter, the output (KNIME table) doesn’t contain the [NULL] rows, but Color Manager sees them?
That’s not clear to me.
Hi @ReWi,
Additionally to the data, KNIME stores certain metadata for each column. For example, if the column is nominal and contains less that 60 unique values, all values that occur in it. If the column is numerical, min and max are stored. The color manager uses that metadata for its dialog. I believe that the Row Filter does not update the domain because in some cases it is good to know that other values “could” exist in the column.
Kind regards
Alexander