Re: continuation and multi-threading

[Keisuke Nishida <kxn30 at po dot cwru dot edu>]

Mikael Djurfeldt <mdj@mdj.nada.kth.se> writes:

> > Because that way you can get continuations to perform -fast-. :)
> 
> I'm all for fast continuations and see their value.  I said don't ask
> me only because I don't know anything about Keisuke's VM---*yet* ;-)
> 
> As I said in a previous response to Keisuke, in some VM designs, it is
> possible to do something which is more efficient than stack copying.
> 
> Do you understand Keisuke's VM well enough to say that breaking the
> stack into frames would be a viable solution?  (I ask this because I'm
> interested, not because I want to put you on slippery ground.)

I understand my VM well :), and I guess fast continuation is feasible.
My VM uses its own stack and the VM itself is a big single C function.
That is, setjmp/longjmp no longer cost too much, since the C stack
won't increase so much after entering a VM.

So, reducing the time of stack copying will be significant, I guess.
I'm not going to implement fast continuations right now, but it is
possible to fix it later, I guess..