Regression in 5.4 beta: Korean CP949 (EUC-KR) SMI/SRT files open as Greek (1253) mojibake. Works in 5.3.0.

[minsk999]

Cause: AutoDetectWordsGreek was expanded with many 2-letter words (να, το, τα, με, σε, μη, θα, τι, αν, ή...). CP949 bytes decoded as 1253 yield random Greek letters that hit these short words by chance, so the Greek check in DetectAnsiEncoding (count > 5, evaluated first) wins before the Korean (949) check is reached. The new "greekCount > cyrillicCount" guard only covers Cyrillic, not CJK.

Also: Settings > General > Default encoding has no effect on open — Se.Settings.General.DefaultEncoding is never copied to Configuration.Settings.General.DefaultEncoding, which GetEncodingFromFile reads.

Workaround: stay on 5.3.0, or File > Import > Subtitle with manually chosen encoding (949).