At 13:49 -0500 12/22/03, Josh Wilmes wrote:
At 12:23 on 12/22/2003 EST, Dan Sugalski <[EMAIL PROTECTED]> wrote: > Second, we're assuming that the *non* threaded case is the commoncase. (This is definitely the assumption that I'm most expecting to> regret in the future) I think it might be good to get started on regretting that as soon as possible ;-)
I think assuming the *non* threaded case to be the most common case, and optimizing for that, is a bad decision.
In that respect, I think we need to think out of the box and take a leap of faith here.
If you realize that most systems out there have threads integrated into the core of the OS, I think it is a very *nix centric assumption that single threaded is the most common occurance.
I think that if you want to have an ultra-fast single threaded program, you would program it in C.
If you want a program that you want to be able to run anywhere using a fast virtual machine, you want to use Parrot. And I think Parrot should come with threads (or whatever you would call program controlled multi-processing) as an integral part of the system.
It's easier to fake fork with threads, than it is to fake threads with fork. Trust me, I've been there...
If the JIT compiler finds out that no threads are being used, it can optimize for that.
Liz
