Caret display uses byte-based offset incorrectly

[davidhewitt]

🚩 Caret display uses byte-based columns — may misalign for non-ASCII on same line

With CodeLoc columns now consistently byte-based, the caret display logic in crates/monty/src/exception_public.rs:228-238 computes positions using byte offsets (line.len(), trimmed.len(), self.start.column). For lines containing multi-byte UTF-8 characters before the error location, " ".repeat(caret_start) generates ASCII spaces based on byte counts rather than display width, which can cause visual misalignment. This is a pre-existing issue (the old mixed byte/character computation was also wrong for non-ASCII) and is not introduced by this PR. Fixing it properly would require a display-width-aware calculation.

(Refers to lines 228-238)

---

Was this helpful? React with 👍 or 👎 to provide feedback.

Originally posted by @devin-ai-integration[bot] in #342 (comment)

[davidhewitt]

NB #342 made CodeLoc character based in the end.