Today I had a most horrible database to deal with, and of course I use KNIME for that - to make it less horrible, of course. :-) Anyhow, the cell splitter node has shown some weird behaviour. Imagine the following, terrible colum:
Decision [this is my column ID]
date 1 (explanation 1)
date 2 (explanation 2)
date 3 (explanation 2)
An apparent mis-use of a free-text field. But anyway, initially not a problem, since we have the cell-splitter node. So splitting at delimiter sequence " (" would do the (first) trick, I thought. And it does. But much later in my pipline I came to realise that the "missing value" fields with a "?" had been split into two "no value" fields with nothing in them - not even the missing value flag "?". Obviously, I couldn't filter out missing value rows by any means after the split, and had to do it before instead.
Question of questions: Bug or feature? Obviously, empty fields cannot be split by any delimiter, so they aren't touched. But "empty cells" not treated as missing values don't feel right, either. What happens, if only the delimiter is missing in some rows, but other content is there? Didn't try it out, it just occurred to me while describing the problem.
Would be grateful for some guidance in this... :-)