On Wed, 2002-03-13 at 10:52, Peter Jay Salzman wrote: > looking through the man pages, i found that signal returns a type > sighandler_t: > > sighandler_t signal(int signum, sighandler_t handler);
Actually, no it doesn't. If you look more carefully, the manpage only uses sighandler_t (an arbitrary type name) to explain the actual prototype of signal(): void (*signal(int signum, void (*sighandler)(int)))(int) Trying to use a sighandler_t or, especially, __sighandler_t, is ill-advised.... (and completely unlikely to change code behavior). As to why the code is failing.... as far as ANSI C is concerned, an implementation doesn't have to signal at all, regardless of what kind of floating point exception occurs. I don't know enough about the specifics of Linux on Intel to tell you whether some circumstances might actually cause the signal to be generated, but I believe someone here mentioned trying integer zero-divide (SIGFPE doesn't have to be floating point, really). Micah _______________________________________________ vox-tech mailing list [EMAIL PROTECTED] http://lists.lugod.org/mailman/listinfo/vox-tech
