@jens @ireneista @misty @bob What I'm saying is that the on disk representation doesn't matter. It doesn't matter that on FAT filesystems the data is stored as UTF-16. When you mount it on a modern unix system, you access it thru the POSIX fs interfaces with the names exposed as UTF-8. When you access it from Windows API, the names are exposed as UTF-16. A wacky hypothetical DSP RTOS might expose them as UTF-32. The important thing is that it faithfully represent a sequence of Unicode characters not a particular representation thereof.