I found that on ARM/glibc 2.5/glib 2.12.something test in strip_unicode() doesn't
really catch invalid utf-8 chars. Also, let's tell user where problem lies straight.

Paul Sokolovsky <pmiscml@gmail.com>
--- a/src/thread.c.org	2007-12-24 00:10:15.000000000 +0200
+++ a/src/thread.c	2008-01-04 22:04:30.000000000 +0200
@@ -349,13 +349,11 @@
 					//g_idle_add(set_media_info, idledata);
 					utf8name = g_locale_to_utf8(parse[name],-1, NULL, NULL,NULL);
 					if (utf8name == NULL) {
-						strip_unicode(parse[name],strlen(parse[name]));
-						utf8name = g_strdup(parse[name]);
+						utf8name = g_strdup("<cannot convert to utf-8>");
 					}
 					utf8artist = g_locale_to_utf8(parse[artist],-1, NULL, NULL,NULL);
 					if (utf8artist == NULL) {
-						strip_unicode(parse[artist],strlen(parse[artist]));
-						utf8artist = g_strdup(parse[artist]);
+						utf8artist = g_strdup("<cannot convert to utf-8>");
 					}
 					
 					message = g_markup_printf_escaped(_("<small>\n<b>Title:</b>\t%s\n<b>Artist:</b>\t%s\n<b>File:</b>\t%s\n</small>"),utf8name,utf8artist,idledata->info);