We flagged a ZWJ in a user's name as hidden text. It was not hidden: the joiner sat between letters of one script, spelling the word correctly. A real name, reported as an attack. Now screenfence flags a joiner only when neither side spells a word nor joins pictographs.
Was this useful?