Michael Sperberg-McQueen comments correctly that "among the techniques which
don't work well for interchange are... using an existing 8-bit character set's
representation of the characters".
It may be of marginal interest to note that the public domain file transfer
program Kermit uses (since version 3, I think) the ISO8859-1 character set,
and will do its best to translate to ISO8859-1 at the sending end and from
ISO8859-1 at the receiving end. Thus, any character which is common to the
sender's and receiver's character set and to ISO8859-1 should go through
alright, and the particularly horrific examples given will be avoided.
I very much hope that the situation will rapidly progress to the point where
manufacturers will offer ISO8859-1 as standard, and TEI can then recommend
it instead of a restricted ASCII, which admittedly is all that we can be
sure of at the moment.
Ciara/n O/ Duibhi/n.