Ben Pfaff <blp <at> cs.stanford.edu> writes: > This replaces exit(0) by exit(EXIT_SUCCESS), which makes sense > because POSIX says that EXIT_SUCCESS is defined as 0. But does > it always make sense to replace exit(1) by exit(EXIT_FAILURE)? > POSIX does not say that EXIT_FAILURE is always 1, but it does say > that some utilities are supposed to exit with exit status 1 in > some cases (e.g. "grep" when no lines are selected), so wouldn't > this lead to a POSIX violation in the most general case?
Not an issue. About the only other EXIT_FAILURE value out in the wild is -1 (aka 255), but as 255 causes xargs to behave differently, it is just too broken to keep. Therefore, since Mar 2007, gnulib's stdlib module explicitly redefines EXIT_FAILURE to 1 for those rare platforms (Tandem/NSK) that tried to be cute by using something that was not the traditional 1. -- Eric Blake
