This is great, I can think of several places where zlib is the bottleneck and would love to speed them up