Non-ascii embedded filename

David Shaw dshaw at
Sat Apr 22 15:41:54 CEST 2006

On Sat, Apr 15, 2006 at 02:14:06PM +0200, Alain Bench wrote:
> Hello,
>  On Saturday, April 15, 2006 at 2:33:21 -0400, feitao wrote:
> > [embedded filenames] gpg has some problem with some Chinese
> > characters. For example, it interprets '‚S' as '\x8S'
>     Confirmed with GnuPG 1.4.3 on Linux. It seems that in characters
> encoding, all bytes that are between 0x80 and 0x9F are munged. Exactly
> all bytes between 0x80 and 0x8F are replaced by the 3 chars "\x8", and
> all bytes between 0x90 and 0x9F are replaced by the 3 chars "\x9". This
> on a correct locale setup, and on a filesystem accepting those bytes in
> filenames. It seems the embedded filename is stored OK, and the munging
> takes place at the --use-embedded-filename stage.
>     Additionally filenames are stored and extracted as they are, in the
> current charset. This gives another problem when the locale is not the
> same during both operations.

Indeed.  This is fixed for 1.4.4.  Embedded filenames are now UTF-8
encoded on the way in, and UTF-8 decoded on the way out.


More information about the Gnupg-users mailing list