bug involving non-ASCII characters used in uniq

Hi,

I've found that the version of uniq that comes with Redhat 9 doesn't
seem to work when there are non-ASCII characters in the lines
(specifically, I've had identical lines fail to be identified when
they contain characters from the GB2312 character set, simplified
Chinese).  On the other hand, sort -u works fine on these lines.  Is
this a known problem?  I would be happy to get involved in any effort
necessary to fix this problem.

Many thanks,

Roger Levy