I really do not like to ignore signals I hear you, but SIGPIPE is documented in this case, and we definitely don't want to just crash with it, and we're doing the right thing (disconnecting) if we see it.
Seems like the only alternative is the inferior-to-the-user choice of invariably disconnecting ... FWIW, SIG_IGN occurs in several other places, though I'm not sure if any are exactly comparable: cmds.c:172: signal (SIGPIPE, SIG_IGN); cmds.c:1534: old1 = signal (SIGINT, SIG_IGN); cmds.c:1535: old2 = signal (SIGQUIT, SIG_IGN); ftp.c:390: if (abrtflag && oldintr != SIG_IGN) ftp.c:507: if (abrtflag && oldintr != cmdabort && oldintr != SIG_IGN) ftp.c:594: oldintp = signal (SIGPIPE, SIG_IGN); ftp.c:697: oldintp = signal (SIGPIPE, SIG_IGN); ftp.c:925: oldintp = signal (SIGPIPE, SIG_IGN); ftp.c:1105: signal (SIGINT, SIG_IGN); ftp.c:1539: signal (SIGINT, SIG_IGN); Best, Karl _______________________________________________ bug-inetutils mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-inetutils
