[gclist] GCBench "benchmark" problem

Florian Weimer fw@deneb.enyo.de
08 May 2001 00:16:24 +0200


"Boehm, Hans" <hans_boehm@hp.com> writes:

> Thus I propose to "fix" it by changing the final test to
> 
>  longLivedTree.left.right == null

Why can't you keep a copy of longLivedTree around in a volatile memory
location?  Isn't this a fix instead of "fix"? ;-)