🚀 go-pugleaf

In article <1992Nov15.134100.13188@ntuix.ntu.ac.sg> eoahmad@ntuix.ntu.ac.sg (Othman Ahmad) writes:
>Scott Dorsey (kludge@grissom.larc.nasa.gov) wrote:
>: 
>: It's possible to build a machine with a single instruction (subtract and
>: skip if equal), but definitely not practical.  I believe a student at U.
>                      ^^^^^^^^^^
>YOu can't say that. Based on my current thinking this is the future of a new
>class of SUPER INTELLIGENT machines, that cannot crash. After all , there is
>only one instruction(conditional).
>	I've a strange feeling this is the way our mind works. 

Make a better idiot-proof box and we'll make a better idiot.  Seriously, do
you think computers always hang and crash because of an illegal instruction?
There's way too much involved in the scheduling, the OS, etc. that even if
you can't execute an illegal instruction, the machine will still choke on
bad code.  Besides which, what happens when somebody makes a coding mistake,
and branches to the wrong place?  The machine will start executing the wrong
instructions, perhaps not instructions at all, and it's behavior then is
undefined.  To sum up, bad instructions  come from the programmer (or compiler,
or whatever is between the programmer's mind and the machine code) not the
instruction set.  There will always be bugs.  Not to say that we can't
minimize them, but show me "crash-proof" and I'll show you an ill-tested
system.

As for the philosophy, who knows, you may be right.

Ted