On Wed, Aug 29, 2001 at 06:40:51PM +0200, Sven Neumann wrote: > > Do you have any good arguments against UTF-8 encoded po-files? > Yes, they are a nuisance. As late as yesterday I had problems with UTF-8 po-files. It was a new textdomain from mandrakesoft, for mandrakesoft online, and it was stated as utf-8 but recode could not handle it, there were some undefined errors in it. Fortunately I just had to translate everything anew and then I could manually set the charset to iso-8859-1. I have had the same problem with some kde po-files, and this was coming from other translators that was not aware of the utf-8 encoding, but just thought that it was iso-8859-1 and edited it as such. I am quite experienced as a translator and have done a lot of work on gnome po files, and even for me utf-8 is a pain in the a** for the time being. For less experienced translators it will just confuse them evenmore. Don't get me wrong - I am all for ISO 10646, all my i18n work has been based on that standard for the last 10 years, and I am even a coeditor of the ISO 10646 standard. And I am very officially for using utf-8 everywhere, both the Internet policies RFC and the POSIX recommendations have my name written all over them. It is just not time yet, and it is not needed as gettext will automatically do the conversion via brunos iconv(). Kind regards Keld
