Virtual Machines

[Arthur A. Gleckler]

|  To: lispos@math.gatech.edu
|  Subject: Virtual Machines
|  Date: Mon, 28 Apr 1997 12:17:34 -0700
|  From: Kelly Murray <kem@Franz.COM>

|  All this talk of a VM is pretty much useless for my goals,
|  which I restate: "to build a platform where Lisp programmers can be
|  successful in the marketplace".

...

|  So I say drop this whole VM discussion.

I couldn't agree more.

I worked in the MIT Scheme project for several years, and one of great
features of that system was that your code was highly portable from
one machine and operating system to another.  You would write a
program on one system (Unix, VMS, Windows NT, and i386, MIPS, VAX,
Alpha, and HP-PA), then recompile on another and it would just work.
(In fact, the same i386 binaries are used in the MIT Scheme runtime on
Unix i386 systems as are used in the runtime on Windows NT i386
systems.)  It was the runtime's APIs that made this possible, not an
interpreter in the VM.

Having to recompile your code when switching systems is not the
problem.  The real problem is having to change your program to fit the
new system.  There's no need to build an interpreted VM or a VM with a
JIT compiler for Lisp portability, and the drawback of using such a
system is potentially dramatically reduced performance.