Script Editor documentation states that a script may contain any UTF-8 characters (Chinese, Russian, ...). This works well on UTF-8 locales.
When a test script with such characters gets saved through GUI on other than UTF-8 locale, for example on Windows (which typically use a CP-xxx code page), the characters get crippled and when the editor gets reopened, they display as question marks.
When a script file is being saved, the GUI doesn't specify the character set in the Java I/O call. The file then gets saved in the default system encoding which may or may not be UTF-8.
The code has been fixed to force UTF-8 encoding on save and open actions. Testing on Windows has proven that it behaves correctly. The fix has been committed to CVS and targeted for 2.0.3.
Fix delivered in 2.0.3 on 29 March 2010.