Re: Re: My arguments

On 11/05/04 Larry Evans wrote:
> >Tracking the pointers in the heap is very easy: the issue is with
> >tracking them on the stack/registers. There are papers, though, about
> 
> Do you mean tracking the pointers conservatively is easy or tracking
> them precisely is easy?  To do this precisely, wouldn't you have to

I was talking about tracking the pointers in the stack or registers
of a thread. Doing it precisely for the heap is easy. Doing it for the stack
is much harder, since it needs the jit to properly report them
and we need also a solution to handle internal calls, where objects are
manipulated by C code.

> know the pointee's type?  Would this information be provided by
> a method stored in the virtual function table of the pointee?  Then,

The information would be stored together with the jitted code
and we can find it simply by using the instruction pointer.

lupus

-- 
-----------------------------------------------------------------
[email protected]                                     debian/rules
[email protected]                             Monkeys do it better
_______________________________________________
Mono-gc-list maillist  -  [email protected]
http://lists.ximian.com/mailman/listinfo/mono-gc-list