Hi, > pocketlinux' version of kaffe had seen more internal > development from Transvirtual, and should contain some > improvements in some areas (AWT, rudimentary jit4, > gc-interface to boehm, more ports & threading systems, > better gcj support ...), but it has been surpassed in > others ;) > > In short, volunteers are needed to merge in the good > bits. > I see. I would like to ask a stupid question: where can I find PocketLinux Kaffe. The URL http://www.pocketlinux.com cannot be reached.
> > Do > > you mean that the GC in > > PocketLinux Kaffe can be plugged into Kaffe as well? > > Only if the gc interface hasn't diverged much between > the two versions. I haven't looked at it seriously > yet, so I can't tell. It would be nice if someone > could try to merge it in, so we can know for sure ;) > Even if the GC interface doesn't diverge much between the two, I wonder if the GC for PocketLinux is suitable for Kaffe. PocketLinux is developed for hand-held or mobile devices. GC for these resource constraint environments cannot be very complicated and has many restrictions. So I would like to take a look on PocketLinux's source code, see how it can have Boehm's GC merged. If possible, I would like to merge PocketLinux's GC into Kaffe (if that's not too tedious), see if there is any performance improvement. Final question: I would like to know if there is any write barrier implemented in Kaffe-1.0.7 currently? It seems that it is somehow implemented in Kaffe (GC_WRITE?), but I cannot find its implementation. Would you mind tell me where does its source code locate? I think a write barrier is important for an incremental/asynchronous GC. Actually, I use Kaffe for my research purpose. Currently I try to study and understand the entire memory management system of Kaffe. I would be very grateful if you can answer my questions. Thanks! Regards, Zoe _______________________________________________ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
