On May 23, 2:04 am, "James Whitehead" <somewh...@overtherainbow.com>
wrote:
"*Anarcissie*" <anarcis...@gmail.com> wrote in message
Derrida - its difference is what is needed for bugs.
The simplest program in assembler would i think be
HLT
No, that might do something.
Hence 'Program'
No, a program is a sequence of instructions, not
something that does something.
You probably mean NOP,
No - i mean HLT - its why i typed HLT.
NOP is not a program - but does do something - yes.
If a particular machine has a NOP instruction, then
one NOP is a program. It does do one thing at least,
advances the sequence counter. HLT does even more;
it changes the run state of the processor.
but even that might do something. The proper null
program has _zero_ instructions.
def Not a program.
See above.
'Not a program' could be random 1s 0s (AKA MS Windows)
That would be begging questions of interpretation.
If there is only one empty set and one null string, as
mathematicians seem to believe, then there is only one
null programl.
I believe the surreal numbers (see Wikipedia) must be
used to count the number of bugs in the null program.
A bug is said to exist when a program does something
it is not supposed to do, or does not do something it
is supposed to do.
Or does something the programer told it too from the analysts mistaken idea
of what the user wanted - but in fact didnt. Etc Etc.
That is not exactly a program bug. One useful term for
such items is "misfeature".
However, because a null program
does not do anything at all, it cannot be determined
for purposes of bug-counting whether it has a bug
or not, no matter what the specifications say.
You probably don't need assembler to write the
null program.
Oh we need it badly - instantiation of objects and polymorphism is modern
day metaphysics...
why divide when you can shift right!
void main() { ; }
ought to do it (in C).
But maybe the ';' is superfluous....
But you have no idea of how much machine code that would generate!
Your faith lies in the writers of the compiler........and there be bugs no
doubt a plenty.
We're talking about the _program_, not its implementation or
execution. Long ago my colleagues and I used to play a game
with IBM's PL/1 compiler, the point of the game being to generate
the largest amount of code with the smallest source. I found
that calling the substring function recursively on both sides of
an assignment had truly astonishing results. The null program
(32K of object code) was not permitted by the rules as too
obvious and easy, or it would have won hands down. You can't
beat 32K / 0 -- or you can, but it would take a pile of math
using transfinite numbers to do it, and a theoretical argument
too, I imagine.
.
