I was playing with localized info files with maxima and have run into
an issue. With cmucl, I do
LANGUAGE=es_ES: maxima-Local
When I try "? build_info", the output is wrong. It contains funny
characters representing U+FFFD (unicode replacement character). This
also fails on clisp and sbcl with errors.
I think the problem is that my cmucl init sets the default encoding to
:utf8 and maxima's READ-INFO-TEXT just opens the file without setting
an external format.
AIUI, we basically just suck in a bunch of bytes and then send them
out directly because the info file has the text encoded using the
desired encoding. If this is the case, I think READ-INFO-TEXT should
probably specify an external format of iso8859-1 so that all byte
values are valid. Or we need to change it to read bytes instead of
characters, and also change DISPLAY-ITEMS to output bytes instead of
characters.
It would probably be most portable to use bytes instead of characters
so we don't have to deal with external formats at all.
Does this make sense?
Surely this is working for other people as it is today, though. (It
seems that this used to work for me as well, but I don't remember
exactly anymore.)
Ray