I believe the point is that the value columns do not have a determined data type. I tried to force a conversion to string but to no avail.
First, I tried to use date/time to string conversion remembering to have date/time fields in the tables, but as unknown is not date/time unsurpsingly I could not select the value columns for conversion.
Then I tried to use Column Auto Type Cast but I did not help either as neither there I was able to select the value columns.
I thought to do the date/time to string conversion before the data enters the diff finder but then I would loose the specs comparison, unless I do it in two diffs making my workflow even messier.
Have you tried string manipulation or column Expression and just add an empty string („“)? Otherwise could you provide a sample or table where this sort of column would appear?
I did not try anything but the conversion attemts. But thanks for the hint. I shall try.
I am not sure what mean with “sample or table”. In any case, I put together a simplified but expanded version of my original workflow also presenting the problem. Please find it attached. table_diff_to_csv.knwf (68.3 KB)
Please note that I removed the Oracle connectors. They will not work on your system anyway if our company did the firewalling right and I a bit secretive too. To reattach connectors, follow the following pictures.