Thanks for your answer. How can I report this to Qlik?
However, this is part of the problem, because if the Boxplot implies always a compulsory dimension, this is going to 'implode' if you choose an occurrence of that dimension. I think a boxplot should be independent by dimension in its 'pure' form, like a histogram (and it is weird, because in Qlik an histogram has a dimension as compulsory, not a measure).
No no, you were extremely helpful, and I edited the post, putting the names of people in the post you have hinted, also I asked to partecipate do the bug report group (as reported on the post): my message was that I tried "the public way" despite to try the "private way", not that you hinted something different. Sorry if my message was misunderstood, you know, reading without the voice make messages misunderstood!