2. Open the Find Files dialog, enter some search text which starts with one or more spaces and contains Unicode characters missing in your current locale. For example, " ﻼﻻﺏﺑ 弋 ф ö " without quotes.
3. Start the search, stop it, close the dialog.
4. Open the Find Files dialog again, open the drop-down list of the Find text field.
5. Instead of the Unicode characters there is a UTF-8 byte sequence shown in ANSI charset. E.g. on my Russian locale it is shown as
(the text field contents however remains correct, only the drop-down list shows incorrect text)п»ї п»јп»»пєЏпє‘ еј‹ С„ Г¶
The problem is that in wincmd.ini this string is stored as UTF-8 with BOM prefix, but because of space the string is wrapped in double quotes, so it seems that TC fails to recognize the BOM signature and treats the text line as ANSI instead of UTF-8.