Hi Charles, Charles Manning <[EMAIL PROTECTED]> wrote: > On Wednesday 17 January 2007 10:15, Juergen Stuber wrote: > > > > I figured out some years ago that fully real-time GC could be done > > with an overhead of two pointers per object and a kB or two of code, > > but that the RCX had too little memory to make it worthwhile. > > I think I have a basic grip of the idea here: Instead of > pointers we have a table of pointers accessed via > handles. Thus the pointers in the table can be fiddled > with without hurting the objects.
no, it was based on the treadmill described in Paul R. Wilson, Uniprocessor Garbage Collection Techniques, and it doesn't move objects around. It doesn't do is compaction, the idea is to get away by just using best fit for allocation. Apparently that is sufficient to prevent fragmentation most of the time. But it would be nice for people to do some reseach themselves, maybe there is something better for the NXT. Now that there is much more memory the trade-offs may be somewhat different. Jürgen -- Jürgen Stuber <[EMAIL PROTECTED]> http://www.jstuber.net/ gnupg key fingerprint = 2767 CA3C 5680 58BA 9A91 23D9 BED6 9A7A AF9E 68B4 ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Lejos-discussion mailing list Lejos-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lejos-discussion
