From: Rick DeNatale <rick.denatale@...>
Date: 2007-12-31T02:46:58+09:00
Subject: Re: Destroying an Object

On Dec 30, 2007 1:44 AM, Sam Smoot <ssmoot@gmail.com> wrote:
> On Dec 27, 12:54pm, Ryan Davis <ryand-r...@zenspider.com> wrote:
> > On Dec 27, 2007, at 07:21 , Rick DeNatale wrote:
> >
> > > It's a general feature of most GC algorithms, not just conservative*
> > > ones.
> >
> > It isn't a (usually) failure of GC algorithms, it is (usually) a
> > failure of GC implementations.
> >
> > My GC didn't have this problem. :P
> >
>
> Neither does .NET's GC IIRC. Which is why the posts interested me. ;)

So both of these GC implementations return the space used by every
object at the instance the last reference to that object is lost?

I've personally not seen a GC like that.  Most delay until the space
is needed.  Reference counting does return objects to the free-list
immediately when the reference count goes to zero, but are
inefficient, and fail to reclaim objects which are in circular lists,
and most have some limit on the reference count which means that
objects whose reference counts overflow get stuck as well.


-- 
Rick DeNatale

My blog on Ruby
http://talklikeaduck.denhaven2.com/