> Hello, > > It seems again, someone else is trying to reinvent the wheel, well the > wheel that nekovm is already: http://www.mozilla.org/projects/tamarin/ > Adobe has contributed a virtual machine (JIT), and a garbage collector > to the mozilla project. > > I was wondering though, if the contributed garbage collector would have > any use for neko (lxr: > http://lxr.mozilla.org/mozilla/source/js/tamarin/MMgc/). It is described > as being a "conservative, incremental" garbage collector. > > Thanks
Thanks for the link Paul, it looks interesting but I think that if a new GC needs to be used for Neko, then it has to be Neko-specific (meaning : Neko-optimized as well). I've already started writing a small GC which works on Win32 but doesn't perform so well right now compared to BoehmGC. I'm planning to continue working on it in the future, but the current GC offers enough performances. As for Tamarin, I'm curious how Mozilla will be able to integrate this in SpiderMonkey, given that's basically the whole Flash9 AVM2, expect Flash-specific APIs. It's indeed interesting to see the complete VM sources being available, especially since I've been reverse-engineering the Flash9 bytecode some months ago when working on haXe/Flash9 support. Now it could be possible to document more accurently the bytecode. OTOH I hope that the VM is perfectly safe or such source could be used to develop Flash-virus which would be bad news. Nicolas -- Neko : One VM to run them all (http://nekovm.org)
