A question for X11/XIM gurus

me> Maybe Xlib checks for LANG=C and says "I don't need no me> steenkin' input method" if it finds it. I don't have X11R6.3 me> source easily available at the moment to check though. I lied :-) I forgot that I built X during the "Sparc/Linux hates XEmacs" fiasco, and the source was still hanging around (another miracle of modern mass storage pricing). sb> I've traced down some locale-related coredumps to the sb> following code in _XimPreConnectionIM (xc/lib/X11/imDefIm.c) sb> when LANG=C and the *inputMethod: X resource is set. sb> /* locale name check */ sb> _XGetLCValues(lcd, XlcNLanguage, &language, XlcNTerritory, &territory, sb> XlcNCodeset, &codeset, NULL); The call above uses functions from at least two other files in xc/lib/X11 :-/, and several of the calls are through method pointers. The pointers have names like `decode_data'; I have no idea how to predict what that call will return for lcd->territory == NULL. (Well, I do have one idea but I'm out of time tonight.) sb> So why doesn't XEmacs drop core when linked against XIM/XLib sb> (it does with Lesstif)? Both commercial Motif (as linked with sb> Netscape and Java) and Lesstif blow chunks and die. My current guess is that for string data those "decoders" substitute a pointer to a null string for a NULL pointer. And that Motif uses its own variant of _XGetLCValues (or something that it calls, or one of the methods) that isn't so careful. -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +1 (298) 53-5091