mattkrick / fast-bitset Goto Github PK

There aren't any tests for the alternate param in the constructor. I can add these; I just wanted to make an issue to document the task.

Any plans to add these?

The docs describe dehydrate as...

Turn the bitset into a comma separated string that skips trailing 0 words...

The tests don't seem to be skipping them, unless I'm misinterpreting what "skip" means?

I would have expected

expect(bs.dehydrate()).toBe('0,1,99');
expect(bs.dehydrate()).toBe('2147483647,99');

based on the README.

Am I just misunderstanding? :)

Allo! When do you plan to cut the next release? Need any help with it?

Also, are you on IRC/Gitter/anything? I was literally going to write this exact library last night, then I found this on npm just a few hours after you published v1.0.0.

It would be really nice to have the .not function for the negation of two bit sets:
111 NOT 010 = 101

Best,
Ctibor

It says on the front page that bit operations are done in O(1) time. Shouldn't that say O(n) time? The time complexity of bitwise operations between two words might be O(1), but this iterates over multiple words, so the time complexity is still scaling O(n/32) = O(n).

This would be different if these operations were done in parallel of course, which they definitely could be.

The latest version of V8 can optimize functions that have let and const. This should provide some performance gain. Node 4.2+ takes it just fine, as does Chrome & Firefox nightly. Still waiting on Firefox & webkit to catch up.