I’ve been rewriting the heck out of my roaring bitmap project lately trying to make it faster. And for the first time its even or slightly faster than CRoaring on all 3 of the benchmarks i’ve shared in my project thread 
.
Still a long way to go before it can do everything CRoaring can. Right now I’ve implemented around 68% of the roaring_bitmap prefixed API methods. So if you need a fast, compressed bitmap which can do logical operations (such as and, or, xor, andnot, flip…), give this project a try.
I’ve been regularly fuzzing it against CRoaring with the zig fuzzer and AFL. And also running “allocation failures” tests. So it should be pretty well behaved. Docs are available although they do need some work 
.