So you're developing a cross-platform library that deals with filenames as a const char * (or the equivalent thereof)? On Windows, please treat the the filenames as UTF-8 and internally convert them to UTF-16 to call the Unicode Windows APIs or only have a wchar_t* function available.
DO NOT pass the filename as-is to the ANSI Windows APIs. With char * being treated as UTF-8 encoded text, this appears to work until an umlaut or another non-ASCII character appears in a filename.
@karotte yes; the older *A functions (e.g. CreateFileA) expect a Windows-1252 (ISO 8859-1 superset) string. that codepage does contain many accented characters, but it is not byte-compatible with UTF-8 so passing UTF-8 strings in will bite you in the ass.
use MultiByteToWideChar to convert UTF-8 to UTF-16, then call the *W suffixed API rather than the *A suffixed API.
Add comment