Re: [dnssd] WGLC on draft-ietf-dnssd-privacy-01

[Christian Huitema <[e-mail]>]

On 6/26/2017 11:41 AM, Stephane Bortzmeyer wrote:

> On Mon, Jun 26, 2017 at 07:18:19AM -0700,
>  Christian Huitema <[e-mail]> wrote 
>  a message of 58 lines which said:
>
>> The solution requires that the participating devices have "good
>> enough" clocks
> Which, IMHO, should be written in the RFC.
>
>> -- to the minute, in practice.
> It is not sufficient. The current text says "We will thus use this 24
> bit number as nonce, represented as 3 octets." If two machines have
> almost perfectly synched clocks, one being at 20:35:44 today, and the
> other at 20:35:43, the values won't have the same first 24 bits
> (1011001010100010101010000000000 vs. 1011001010100010101001111111111).
>
> There is no obvious solution. We cannot have "fuzzy" comparisons with
> nonces.
In fact there is a solution, which I implemented in my tests. Given a
clock precision X, the device knows that its real clock could be
somewhere between T-X and T+X. There are a finite number of time stamps
in that interval, and it should be ready to accept them all. Or query
them all, if it is doing polling.

-- Christian Huitema