Having serious doubts about claims being made about low-level performance and its relation to method size in the LMAX #qcon talk