Re: [h-e-w] unicode path displays wrong in minibuffer

[Eli Zaretskii]

> Date: Tue, 29 May 2007 09:07:06 +0200
> From: "Johan Rex" <address@hidden>
> 
> I just thought Windows uses unicode for filenames.

It does, but it almost never returns you a file name encoded in
Unicode.

> Anyway, I think I know how to reproduce it. I have deleted my .emacs file
> and .emacs.d directory so as to rule that out. I have .txt files associated
> with "emacsclientw.exe -n". When I doubleclick on a .txt file with swedish
> characters in the filename e.g. "bäst.txt" the error occurs. But when I
> invoke emacs from the commandline with
> emacsclientw -n "c:\Documents and Settings\23052547\Desktop\b=E4st.txt" all
> works well.
> 
> So something happens when I doubleclick a .txt file that doesn't happen when
> I use the command line.

Ah, clicking! ah, emacsclientw!  I'm quite sure that the culprit is in
how emacsclientw receives the file name from Windows and communicates
it to Emacs, then.  After all, emacsclientw doesn't have at its
disposal the full arsenal of handling non-ASCII characters that Emacs
has, and I'd bet emacsclient is not written in Unicode-clean way
(which on Windows would require to use wchar_t or the TCHAR trick).
Just a guess, for now, so if someone is willing to dig into this,
please feel free.