> It isn't - I incorrectly assumed that fcntl(fd, F_DUPFD, n) is > semantically equivalent to dup2(fd, n), but it isn't. I'll have a > patch committed RSN. do we get a refund? :-) randy