Combine these two together, and it paints the picture that they don't feel
confident they can meet their performance targets with tried-and-tested
well-established "rusty" GC techniques. 

 

 

Even with local pointers they are struggling to get past java performance (
which is not bad)  so they are rolling stuff back first to go is segmented
stacks .. ( Another idea used in singularity , but for singularity the
bartok compiler could do this  much better)  and the stdlib is getting an
overhaul.  They are well aware that a mark and sweep ( which is under
development ) will give much better performance  but they feel this is more
for big apps not stdlibs and systems programing .

 

Ben

 

_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to