|
* Alexander Gladysh: > I guess that the cost of returning function is compensated by proper tail > recursion in the "chain" case, and that x1.09 slowdown of the "plain_chain" > case is that cost itself. I think you are probably measuring GETUPVAL performance.