I would concur that Ruud should decide how to proceed.

I don't mind either way, whether the function is changed or whether the documentation is updated to make people aware of the empty space issue. It's irrelevant whether the function is technically correct and the issue is Data - what is relevant is what programmers are expecting to get when using the function.

Judging by the fact that many experienced programmers are being caught out by this, clearly demonstrates the need to update the documentation or change how the function works.