On Tue, Oct 17, 2006 at 02:14:28PM +0200, Georg Baum wrote: > Am Dienstag, 17. Oktober 2006 14:06 schrieb Abdelrazak Younes: > > Abdelrazak Younes wrote: > > > Hum, no objection but a more correct fix would have been to convert to > > > unicode. > > > > Like the attached. Not the complete port but at least a first step ;-) > > That is unfortunately no conversion to unicode. Lars used this construct > several times, but it is IMO cheating. We should eliminate all occurances > of > > docstring(s.begin(), s.end()) > > where s is a std::string because this generates garbage for nonlatin1 > input. That may explain the occasional "invalid char sequence" messages > from iconv.
It generates garbage for latin1 input, too, when a char is outside the ascii range (i.e., greater than 127). -- Enrico
