At 11:53 AM 11/28/00 +0100, Petr Ferdus wrote:

>Perhaps this could be done more generally.

Well, it depends on how dependent this is of an encoding vector. I think
that in that case we need some specific unicode def section per encoding
vector. I did some tests and i think it can be implemented without
noticable overhead. You may test the attached file [just include it on top
of your doc] and don't include obj's manually. This test hack may also be
of interest to polish etc users i think. Are there more such predefined
lists of unicode mappings? 
Hans