One idea i've got during the JVM Summit,
here is a code that implements the Stream.filter/map/reduce operations using 
the new loop method handle,

Currently, the code that use method handles is far slower than the equivalent,
so the challenge is how to make it perform better.

Maybe, it means we should introduce a special method handle combiner/something 
else that explicit ask for compiling to bytecode + JITing a method handle.

mlvm-dev mailing list

Reply via email to