Radosław Bułat

I have the same problem. I've installed REE by classic ./configure && make install.

**tl;dr** Some benchmarks have other overhead. For example `call vs send vs method_missing` claims that send is only 1.14x slower and method_missing 1.42x slower than direct call. After removing all...

I woulnd't disable it too. The point of my comment was that we shouldn't tell people that sequential assignment is slower than parallel. If we would do that then we...

One more note about this benchmark. When I did this research I also checked how many local variables are used in "average" method (I used `ripper` library to do that)....

Sure, no problem. Should I bump to `0.9.1` or `0.10.0` version?

@jarl-dk Could you review and eventually merge this pull?

On my machine (OSX 10.11) `-O2` actually optimized it away: ``` $ gcc -O2 sum.c $ time ./a.out 500000000 real 0m0.004s user 0m0.001s sys 0m0.002s ``` `-O1` gives the same...