Your micro-benchmark improvements are significant, but do you have
evidence to suggest that the performance of this method is actually
critical to real applications?

In other words, is the added code complexity really worth it?

- Mark

Reply via email to