On Thursday, 3 January 2013 at 18:54:32 UTC, thedeemon wrote:
+1 to better GC as a project. An easier but still possibly interesting goal would be changing current (or better Rainer's) GC to use parallel marking and lazy sweeping, same things that were done for Ruby interpreter recently. That should make GC pauses much shorter and it doesn't require rewriting most of the GC.
I think we'd better look for performance boost in concurrency than parallelism here. Both TL and immutable are very concurrent GC friendly.
