Unicode filenames on Windows

[Lukáš Lalinský]

On Po, 2007-11-05 at 11:27 +0100, Linus Walleij wrote:
> The way it works with other portable libs such as glib is that all filenames
> are given in UTF-8 (char *), then converted inside the lib to wchar_t
> * or whatever
> is needed (native UTF-8 on POSIX systems e.g.). (iconv() does the trick
> for you in any case).

This is what I did in my personal branches, but I don't think it would
be a good idea for TagLib. Filenames on UNIXes are considered binary, so
representing them as unicode is not lossless. I think TagLib is
low-level enough to be able to work with all possible filenames, only
only the ones representable in Unicode.

Besides, such change, even though binary compatible, wouldn't be
functionally compatible. And if we are going to break compatibility, I
think a system with pluggable IO classes would work better (so people
can use gio, kio, QFile, ...).

Lukas