Lambda the Ultimate

inactiveTopic O'Reilly blog entry on battling VMs
started 11/12/2003; 11:15:12 AM - last post 11/17/2003; 2:30:58 PM
Ken Meltsner - O'Reilly blog entry on battling VMs  blueArrow
11/12/2003; 11:15:12 AM (reads: 223, responses: 4)
David Sklar wrote a short article on O'Reilly's blog site about VMs competing for the hearts and minds of "LAMP" developers:

  • PHP# (PHP for the CLR)
  • Java for Parrot

and other odd ideas:

Virtual Machine War

Dan - Re: O'Reilly blog entry on battling VMs  blueArrow
11/13/2003; 10:30:47 AM (reads: 189, responses: 1)
I really ought to post a reply there, but it's hard to work up enough enthusiasm to bother.

The idea that somehow there's going to be one (or a series of) great battle and that there will emerge One True VM is an utter load of crap, almost as big as the One True Language, One True Editor, or One True OS load.

There are more than six billion people on the planet and an insanely dizzying number of devices and pieces of machinery supporting that population, with an ever-growing number of them integrating a computer in one form or another. Anyone who thinks their [insert software or hardware widget here] is the right answer for every situation that calls for whatever class of hardware or software their thing falls into has an ego big enough to be seen from orbit, and desperately needs to get over themselves.

Ehud Lamm - Re: O'Reilly blog entry on battling VMs  blueArrow
11/14/2003; 1:07:47 AM (reads: 168, responses: 0)
...and desperately needs to get over themselves.

Well put!

Patrick Logan - Re: O'Reilly blog entry on battling VMs  blueArrow
11/14/2003; 7:39:23 AM (reads: 151, responses: 1)
But on the currently most popular platform, in a few years, language providers will have to adapt to a system where they either use a VM that may not have been designed for their language, or their language runs in an environment alongside a VM that implements the platform services.

i.e. life on a Windows system in a few years forces a perhaps unwanted VM into your picture.

I think that is a good thing, since it seems that VM will become more generally useful over time. I wish they had layered their architecture better, but that's life.

Dan - Re: O'Reilly blog entry on battling VMs  blueArrow
11/17/2003; 2:30:58 PM (reads: 118, responses: 0)
There's not much you can do if someone holds a gun, real, legal, or metaphorical, to your head and says "This is the way it is, chuckles!" You're screwed that way, whether it's a (currently) mythic .NET-only windows or a more real JVM-only cellphone. A lack of choice doesn't mean mean what you have is the best--all it means is that what you have is what you're stuck with.

I'm pretty sure that no amount of layering, short of a radical rethink of what belongs in kernel/exec/supervisor mode will ever get you layering that means anything. (I'm not sure that'll get you anywhere either, but at least there's a chance) That means moving a chunk of functionality out of the VM and into the OS, but for that to mean anything for alternate VMs that want any chance at performance the crossing has to be relatively inexpensive--not necessarily in terms of mode switch time but rather in the expense of setting up the call. (Fast mode switches don't get you anything if a no-side-effect, pure language has to build up and tear down a set of 'native' objects to do anything)