Hi Keith,

I updated the bitstring implementation to support 64 bit string lengths. The change seems pretty simple, but it touches code I'm nervous about (in btree). Could you do a review and pull if it's OK?

Also, while I was doing this I came came across the gcc __builtin_ffs, which turns into a single processor instruction. Should we think about using that in the _bit_ffs implementation?

Thanks,
Alex