Warning: Missing charsets in String to FontSet conversion

[stephen at xemacs.org]

Manuel Cebrian writes:

  > I am running xemacs on kubuntu dapper drake with the default
  > settings (just installed).

"kubuntu dapper drake" is meaningless to us; if it turns out to
matter, it's by definition an Ubuntu problem, not ours.  (That doesn't
mean interactions with the environment aren't our problem, just that
if it's one particular distro, that distro is going to have to
reconfigure and rebuild its packages---we can't.)

Please use the M-x report-xemacs-bug function to report bugs; it will
tell us what we really need to know: XEmacs version, configuration
options, libc and kernel versions, packages installed and versions,
etc.

 > $ xemacs
 > Warning: Missing charsets in String to FontSet conversion
 > 
 > This message appears allways. There is a ?related? bug when I compile 
 > using xemacs M-x compile, there are some strange (incorrect) caracters 
 > in the compilation window (\200\230):

These bugs are different, but both due to changes in the environment.

XEmacs is based on X11 from 15 years ago, and expects the "legacy"
bitmap fonts.  However, many modern distros don't supply them, so you
get the warning.  Compile with --with-xim=no (XIM is another legacy
from X11R5) and the "missing charsets" warning will go away.  This
will almost certainly never be fixed in the 21.4 series; anybody who
cannot live with --with-xim=no will be told to upgrade to 21.5 for
many reasons.  I expect it will be fixed "en passant" as refactoring
of 21.5 proceeds; it is not sufficiently urgent to devote immediate
effort to it since the --with-xim=no configuration is known to work
around the problem for almost everybody.

The other bug is either in gcc or in a wrapper supplied by Ubuntu; it
is using non-ASCII characters (specifically, typographically correct
balanced single quotes) in the output from the compiler.  IMHO this is
unwise at best.  Even today there are plenty of post-processing
applications that are not prepared to handle anything but plain ASCII
text.  My strong recommendation is that you find out how to defeat
this feature, so that gcc produces the "informal standard" ASCII-only
output.  This is an Ubuntu or GCC question; please direct it to the
appropriate channel(s).

If you are using XEmacs 21.5.6 or later, then you can probably put
`(setq default-process-coding-system '(utf-8 . raw-text))' in your
init file and get pretty output.  However, you will almost surely lose
some or all of the facilities that use functions that parse the error
output, as they depend on matching the ASCII backtick and apostrophe
characters instead of the "beautiful" Unicode left and right single
quote characters.  This will take a very long time to fully fix, as we
can't just do a global replace of "`" with "[`?]".---we need to find
out which occurances need to be changed, and which should not.

HTH

Steve