My software environment: knime 5.3.1. Operating system: win10.
This problem occurs in the Row Filter node in the new version 5.3.1.
Test process description: I want to filter out the row with “Row number = 2” according to “Row number”. The filtering is executed correctly, but when the configuration interface is opened for the second time, the correct setting is not loaded.
That is, I set the parameter to “2”. When I open it for the second time, I hope it is still “2” instead of “1”.
Thank you for your reply. The difference between “Row Index” and “Row Number” as I understand it is that “Row Index” starts from 0 and “Row Number” starts from 1.
But my problem here is not this concept. Instead, I have set “Row Number = 2”. My purpose is to take the second row of the entire table (the second row counted from 1). The filtering is executed successfully and there is no problem. My problem is that I set the parameter to 2, and when I open the configuration interface for the second time, it shows 1.
For another example, I set “Row Number = 3” and the execution is successful. When I open the configuration interface for the second time, I hope it shows 3, but it still shows 1.
The problem is that in the new version 5.3.1, in the configuration interface of the new node, when the interface is opened for the second time, the previous settings are not loaded correctly.
PS: @tomljh to better reflect the seriousness of the discovered bug I’d suggest, if still possible, to alter the title to something like “Row Filter: Setting lost after re-execution”
thank you for reporting this problem and thanks @mwiegand for the nice video. This helps us a lot in understanding the issues and will speed up the development time. We could isolate and fix the problem (internal ticket UIEXT-2110). It will be released with the next version (5.3.2).
Thanks again for reporting and sorry for the inconvinience,
Daniel
Continuing “off the topic” with the “off-by-one” errors theme, the good news re the new Row filter/splitter/cropper nodes is that they have removed the inconsistency in KNIME, mentioned in the post that @mwiegand linked above, as demonstrated by this workflow (KNIME 5.3):