non-ASCII error messages from gcc

[Jakub Jelinek]

On Wed, Sep 28, 2005 at 04:56:09AM -0500, Berna Massingill wrote:
> On FC2 (and earlier RedHat) systems, error messages from gcc seem
> to have been straight ASCII.  On a couple of FC4 systems, output
> includes some non-ASCII characters; specifically, the characters
> used for single quotes (previous ` and ') have been replaced by
> something multi-byte.  (I base this claim on redirecting stderr
> into a file and then displaying file contents with "od -c".)
> The sequence that replaces ' displays okay, but the sequence that
> replaces ` does not (appears as boldface u with umlaut).
> 
> My guess is that there's an environment variable somewhere that's
> not getting set quite right; the man page for gcc seems to be
> indicating that what gets produced for error messages is controlled
> by environment variables LANG and LC_MESSAGES.  LC_MESSAGES is unset;
> LANG is set to en_US.UTF-8.

It is LC_CTYPE that tells the compiler what the display character set is.
If LC_ALL nor LC_CTYPE is present in the environment, ctype defaults
to $LANG (if set), otherwise to C.
If your terminal can't display UTF-8 encoded characters, then I guess
you shouldn't be using UTF-8 character set...

Works just fine here in FC4 both on the console and in gnome-terminal.

	Jakub