4 Apr
1999
4 Apr
'99
3:05 a.m.
>Actually, I thought they had microcode or assmebly
language. Therefore,
>they have a compiler to create the microcode or assembly language. It may
>not be well-documented (or advertised) but it's there.
This is a mistaken impression. Actually, microcode is
not compiled. For
any computer which is microprogrammed, there is a sequence of
microinstructions which implements each executable statement of the
computer. In the case of the Lisp machine (such as those built by
Symbolics - like the 3600), the machine instructions are, in fact, the
operators of Lisp. So, there is within the microcode of a List machine
a sequence of microinstructions which implements directly the function
of a Lisp operator, such as CAR.
I was unclear.
I meant that the microcode/assembly language words _corresponding to a
particular LISP program_ were created from that program and then executed.
If you define a function (like the ever-popular factorial function)
something has to be stored in memory as the definition; presumably it is
some sort of primitive (as in not-easily-decomposed) machine language, and
presumably there is a program that converts source text into object code.
So wouldn't that converter be a compiler? I believe that a number of subtle
details happen during the conversion process, so you couldn't even say the
compiler is a simple compiler.
-- Derek