I am playing with safe signals again. With this scheme C code just sets a pending flag, and then perl level handler is called next time round the runops loop. I have a half remembered worry. I seem to recall some system where if the SIGCHLD signal handlder did not do a wait() of some kind then when it returned SIGCHLD was posted again, and again, and ... The attached program attempts to test for this case. Will those of you with various UNIXes try it please? (Linux+glibc is okay). If I have remembered correctly, then on such awkward systems we need to do a wait() in the handler and save status somewhere. Which is a pain as one does not know (in the general case) the pid in advance to pre-allocate storage. -- Nick Ing-SimmonsThread Next