Simplify usermod.c

[alejandro-colomar]

In parsing the argument to the --expiredate option, there's a special case for "":

shadow/src/usermod.c

Line 1076 in 7687ae4

if ('\0' != *optarg) {

			case 'e':
				if ('\0' != *optarg) {
					user_newexpire = strtoday (optarg);
					if (user_newexpire < -1) {
						fprintf (stderr,
						         _("%s: invalid date '%s'\n"),
						         Prog, optarg);
						exit (E_BAD_ARG);
					}
					user_newexpire *= DAY / SCALE;
				} else {
					user_newexpire = -1;
				}
				eflg = true;
				break;

However, (the new version of) strtoday() already handles that special case:

shadow/libmisc/strtoday.c

Line 77 in 7687ae4

if ((NULL == str) || ('\0' == *str)) {

	if ((NULL == str) || ('\0' == *str)) {
		return -1;
	}

	...

So the calling code could be simplified to the following:

			case 'e':
				user_newexpire = strtoday (optarg);
				if (user_newexpire < -1) {
					fprintf (stderr,
					         _("%s: invalid date '%s'\n"),
					         Prog, optarg);
					exit (E_BAD_ARG);
				}
				user_newexpire *= DAY / SCALE;
				eflg = true;
				break;

---

There are 2 versions of strtoday() (actually 3), and at first I thought that the old ones might not handle correctly the special case. However, I found what could be a bug, which I think would handle "" correctly by accident. Instead of returning -2 on error, the old functions return -1, and since they don't handle "" specially, they will report an error. Since -1 is handled not as an error but as a 'never' by the calling code, it would work.

I'll send a pull request soon and link to this issue.

[ikerexxe]

As the PR was merged I guess you could close this issue. Am I right?

[alejandro-colomar]

Ahh, I forgot. Sure