[comp.arch] Why we can't cater to bad code

dce@mips.UUCP (02/25/87)

Ok, folks, it's time to 'fess up about the NULL pointer deal.

The fact is that we need to have some new obscurity in Unix systems:

	User (looks like his name is Melvin): I think the compiler is
		broken because this program doesn't work on the new machine.
	Guru (looks like Ed Gould; who else?): Let's see...
	User: I do a strcpy() of the result of getlogin() and ...
	Guru: YOU DO WHAT?!? Hahahaha! What? Were you raised in a barn?
		Everyone knows that you can't dereference a NULL pointer!
	User: But...but...but it works on the Vax!
	Guru: That's no reason, silly computer nerd! You are hereby relegated
		to the position of computer elf until 1999 or until you go
		to work on a VMS system.

If everyone could do it, we wouldn't get paid so much (well, not that
much really, Mash).
-- 
			David Elliott

UUCP: 	{decvax,ucbvax,ihnp4}!decwrl!mips!dce, DDD:  	408-720-1700

guy@gorodish.UUCP (02/26/87)

>	Guru: That's no reason, silly computer nerd! You are hereby relegated
>		to the position of computer elf until 1999 or until you go
>		to work on a VMS system.

It is worth noting that if the aforementioned computer elf goes to
work on a VMS system, they'll learn not to dereference null pointers
*real* fast.  VMS does not permit references to location 0 unless you
do a bit of remapping on your own.  (It is also worth noting that,
according to a reliable source, a version of paging VAX UNIX done a
while ago at Bell Labs also did not permit references to location 0.)