GC conservatism -- again

[Walter Bright]

Steven Schveighoffer wrote:
> This doesn't help.  The requirement is that it is one contiguous piece 
> of memory.
> 
> I don't know if we can solve the GC problems (although it seems that is 
> all people want to talk about here), I was wondering if we can provide a 
> library/language supported solution to enable better manual memory 
> management in these extreme circumstances.

I suspect that in general when one has such gigantic allocations, there are only 
a couple of them in a program, and it can be managed manually by using malloc/free.