Clément Renault

> The branch includes several slightly different variations of the same alg. It also checks in the sample datasets into the repo for benchmarking. It's hardly merge ready, but it's...

Wow, the diff with the multi-or is quite impressive, I don't have a comparison with the multi-xor, though! 😃

Thanks a lot @josephglanville, indeed it seems like I didn't take enough time to make this method work. I hope you fix it, and wait for your PR, thank you...

One simple way for a good enough `impl Hash for RoaringBitmap` could be to iter on the numbers. The `impl Hash for Treemap` could simply be a basic derive as...

Thank you for your issue. I have done some research related to this area that you can read in the issue https://github.com/RoaringBitmap/RoaringFormatSpec/issues/9. I don't know if we want to introduce...

Hey @Nemo157, I totally understand what you proposed! I started working on the "hardest" part of the problem already and found out that this "multi-Pairs" type is not **that** easy...

> In addition to the above we can get rid of `html-reports` feature of criterion and bench will have much few dependencies What do you mean by _get rid of...

> This being said, what you describe is also undefined behavior in C/C++ in the sense pointer casting sense, so don't do that. In C++, what you do is that...

> Note that both of these solutions (Go and Java) are used in production systems. > > I don't think that the issues of value alignment are relevant here as...

> How confident are you about this statement? The frozen serialization and deserialization in C are tested with sanitizers on. With sanitizers, undefined behaviour, including unaligned reads, are usually identified....