two co-existing projects
Tue, 29 Apr 1997 00:16:53 -0500
> Forgive me my ignorance. I am watching this discussion with
> fascination but there are a lot of thing I do not understand. I am not
> a computer scientist but just a physics graduate student. In relation
> to Lisp --- I am just a satisfied customer. Could anyone explain me
> what VM is good for? Is it just a tribute to Java hype? Does anyone
> seriously hope to replace Java VM with New(TM) and Improved(TM)
> Virtual(TM) Machine(TM)?
Virtual Machines come in many flavors - what confuses the issue is that many
people confuse the *definition* of a VM with its *implementation*. If a VM is
defined at too low a level there's not all that much you can efficiently do
to boost performance since most code transformations and optimizations are
unavailable/unproductive. My particular interest is in a VM that defines a
portable interface for compilation/translation of dynamic high level languages
into a compact representation which can then be used for real time native machine code
generation (NOT that JustInTime crap) on any machine/processor/architecture. Nothing
is being interpreted at runtime. You -could- interpret this VM code if you insist
(should result in a 2X to 4X speedup over most bytecode interpreters by itself)
but why stop there when you are so close?
As for the JVM -- phooey!
More than one VM can run at the same time you know, even the grungy interpreted ones.