# Regression test for a bug where tvix would crash in nix_escape_string # because it counted the string position by unicode code point count, # but then used it as a byte index for slicing. Consequently, it would # try slicing 💭 in half, thinking the first element to be escaped was # at byte index 2 (i.e. the quote). "💭(\":thonking:\")"