[texhax] Bug#446476: natbib cannot handle utf8

[Ralf Stubner]

On Sat, Oct 13, 2007 at 12:24 +0100, martin f krafft wrote:

> So I tried *again* to remove the utf8 characters from the bibtex
> entries, and this time I guess I got them all (and previously must
> have forgotten one), and suddenly it works.

recode --diacritics UTF-8..LaTeX

> But this can't be, for I *surely* have used UTF8 in bibfiles before,
> and this is not really the same as http://bugs.debian.org/382075
> where the problem was that a ü character ended up being used as the
> key for an entry.

There the first byte from the UTF-8 representation of 'ü' was used in
the key, which triggered an error. With natbib the full author field
ends up in the key (see main.bbl), so this particular problem does not
happen. 
 
> So this is a natbib bug, it seems.

It might also be a problem with ucs.sty's implementation of UTF-8. At
least your sample document works for me if I use the UTF-8
implementation from inputenc, ie

\usepackage[utf8]{inputenc}

instead of

\usepackage{ucs}
\usepackage{inputenc}
\inputencoding{utf8x}

cheerio
ralf