Roman Leventov
Roman Leventov
There are JNI impls in this comparison already, so it's legitimate. However it will lose competition on small input sizes, because JNI overhead is ~ 50 ns as far as...
I find myself visiting [this SO question](https://stackoverflow.com/questions/1519736/random-shuffling-of-an-array) to copy the code from time to time. This definitely happened more than three times. Avoiding `Collections.shuffle(Ints.asList(a))` is a performance habit.
> However we already integrated findbugs tool in our build process, don't you think it would be better to replace that instead of using two similar tools side-by-side? @anmolnar no,...
@anmolnar see http://errorprone.info/docs/installation, the first paragraph: "Please note that Error Prone must be run on JDK 8 or newer. It can be used to build Java 6 or 7 code...
I didn't have time to consider your latest questions yet
@anmolnar if you try to run `ant test` while the test are not compiled yet (i. e. they have to be compiled), and Ant uses Java 7 which is installed...
Maybe you provide not enough optimization flags for GHC? For this test I have the following `Core`: ``` main3 main3 = \ @ s_a2nv s_a2xF -> case newByteArray# 24 (s_a2xF...
https://github.com/OpenHFT/Zero-Allocation-Hashing/commit/8a6d6a1162f8a704eb3ffc2648fdae7a39885fcd
Yes, I can do it later. Also: probably this function as well as compare_numba_and_py() should use `itertools.zip_longest` rather than `zip`.