Return to dump CVS log

Up to [Research Unix] / researchv9 / cmd / adb / null

researchv9-SUN3(old)

From research!cithep!sickkids!mark Thu Nov  7 15:24 EST 1985
I need to make a couple minor changes to dump, and having looked through
the code I've been reminded of just how sleazy it is.  I've already told
you about how it uses 1 rather than 0 as its normal exit status, thus
ignoring all the standard unix conventions regarding such things.  Now
I found another goodie.  There's an option which permits messages to be
broadcast to all logged in users whom the program deems to be operators.
The broadcasting is done by a subprocess.  (Never mind why; that's another
story.)  What you may be amused and/or disgusted by is the following code
fragment ...

	switch (pid = fork()) {
	case -1:
		return;
	case 0:
		break;
	default:
		while (wait(&s) != pid)
			continue;
		return;
	}

	if (!notify || gp == 0)
		exit(0);

Given that the default is notify==0, think of all those poor processes,
spawned only to die an immediate death!

But wait -- there's more!  It goes on to do its thing.  When it finishes
up, it does ...

	Exit(0);	/* the wait in this same routine will catch this */

... which is correct, provided you think it makes sense to do all this in a
subprocess.  However, before getting to this point, one finds the following:

	if((f_utmp = fopen("/etc/utmp", "r")) == NULL) {
		msg("Cannot open /etc/utmp\n");
		return;
	}

Yes, "return", not "exit"!  Who wrote this mess, anyway?  It seems worse
than even my wildest nightmare of the possible product of a Joy/Horton
collaboration.

This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.