@vfaronov your second bug is interesting (and not really the PRs fault I don't think) if a character is displayable in the editor it should be displayable in the symbols pane, so maybe tagmanager or the parser is mishandling UTF-8 characters so its not actually the same string (though I thought that tagmanager at least had been fixed). And parsers must handle illegal syntax and characters since, as you point out, they can occur when users are typing, but bugs may still exist.
Interesting that its different between Win and Lin, are they running with the same locale? I'm thinking C character classification functions that are locale dependent.