Since I also already had once problems with weird unicode in my source by wildly copy pasting from pdf documents, I would really appreachiate a nice error message about illegal unicode characters. Non ascii whitespace and symbols should be illegal outside of string literals.
The only unicode symbols that I personally think are reasonable to accept would be the following: * `∧` ⇔ `and` * `∨` ⇔ `or` * `¬` ⇔ `not` It would allow a consistent nim source code where every operator consists only of symbols. But that's for another topic.
