Maintaining the correct character set is important. The encoding,
however, is supposed to be changeable. And especially since QP is
strongly discouraged except for some very special circumstances,
translating all incoming messages to 8-bit or Binary encoding is a
good practice. Unless you are connected to a host which still uses
7-bit only (highly unlikely), outgoing messages ought to be in
8-bit encoding. But it is always important to maintain the character
set of the original message when you quote something. This will
of course not always be possible, but at the very least a best-fit
translitteration should be used. In OS/2 it would be best to set the
codepage of the editor task to the one used in the message. Editing
ISO 8859-1 messages in CP 1004 instead of converting to CP 850 will
retain any incorrectly used codepoints from the original message.