Re: Calculation of the hash width in cache-digests

On 18 July 2016 at 18:20, Kazuho Oku <kazuhooku@gmail.com> wrote:
>
> 1. if the url list is empty, set N=0, P=0 and return
> 2. set NP to round_up_to_power_of_two(num_urls / max_false_positive_rate)
> 3. create a list that contains sha256(url) of all the URLs, each
> truncated to log2(NP) bits
> 4. sort the list and remove the duplicates
> 5. set P to round_to_nearest_power_of_two(NP / size-of-the-list)
> 6. set N to NP / P


Yes, I believe that this is exactly what I proposed.

I agree with you that the choice of values for N and P need to be
signaled, but they don't need to be selected in the same way.  The
draft could be clearer about this; even to the point of explicitly
saying that.

Received on Monday, 18 July 2016 16:30:00 UTC