== Quote from Craig Black (craigbla...@cox.net)'s article Does the GC have knowledge of > pointers on both the stack as well as the heap?
Also, probably most of the problem with false pointers is on the heap anyhow. The stack is usually on the order of a few 10s of kilobytes, the static data segment maybe another few 10s of kilobytes, whereas the heap can be hundreds of megabytes. I think that, given how little progress has been made on improving the current GC implementation and how hard a problem it is, adding precise heap scanning to the current GC is a good 80/20 solution until D becomes popular enough to earn hordes of money for research into improving its GC.