Glen Starchman <[EMAIL PROTECTED]>
> No, I am literally meaning having multiple instances of a single VM for
> executing bytecode in parallel.
>
> What I am really trying to do is emulate user-level threading within my
> language without having to rewrite great chunks of its core. My thought is
> that if I spawn a new VM for each user "thread" and then synchronize the
> execution results at some point I could have a very rough emulation of
> threading.
Oh, I see. I've done things like that in my own code (that source is
not publically available, sorry) though only with cooperative
multitasking (which wasn't a problem as I controlled the compiler and
language being compiled too.) How do you propose to handle blocking
syscalls[*]? (Portable) name resolution[**] is among the thorniest
issues ahead... :^(
Aside from that, simulating MT with a cunning VM should be possible; I
believe that is what early versions of Java on Solaris did (try
looking for stuff on how "green threads" versions of Java were done.)
HTH.
Donal.
[* My code ignored the issue as it was doing hardware simulation. ]
[** Talking raw DNS is not portable due to issues with other name
resolution services like NIS and, conceivably, LDAP. ]
--
Donal K. Fellows, Department of Computer Science, University of Manchester, UK.
(work) [EMAIL PROTECTED] Tel: +44-161-275-6137 (preferred email addr.)
(home) [EMAIL PROTECTED] Tel: +44-1274-401017 Mobile: +44-7957-298955
http://www.cs.man.ac.uk/~fellowsd/ (Don't quote my .sig; I've seen it before!)
