David Carlisle wrote -- > But whether the internal > canonical form is a unicode number or a latex style 7bit string \'e > the issues of mapping between input encodings and this internal form, > and from there to font encodings, are probably about the same. And I just want to agree with this. In fact I would go a lot further and say that the problems raised in this discussion such as the following are the same whatever system you use to do quality typesetting: what is a character? what is the relationship between character strings and relatively positioned glyphs on a surface? Thus LaTeX and its choice of internally using 7-bit strings is a also a mere detail. And these problems do not go away just because you use larger integers to represent text streams. chris