Re: [hybi] "fresh" and "uniformly at random":

[Denis Lagno <[e-mail]>]

oh, well, so you'd better avoid phrase "not used previously" in the
first place.  It was highly misleading.

On Mon, Jun 20, 2011 at 11:31 AM, Adam Barth <[e-mail]> wrote:
> You can sensibly apply the term fresh to 1-bit values if you like.
> The important aspect is independence from your previous choices.
>
> Adam
>
>
> On Mon, Jun 20, 2011 at 12:26 AM, Denis Lagno <[e-mail]> wrote:
>> maybe I miss something but in the text "fresh" is applied to 32-bit values..
>>
>> On Mon, Jun 20, 2011 at 6:57 AM, Adam Barth <[e-mail]> wrote:
>>> On Sat, Jun 18, 2011 at 12:34 AM, Denis Lagno <[e-mail]> wrote:
>>>> On Sat, Jun 18, 2011 at 10:27 AM, Adam Barth <[e-mail]> wrote:
>>>>> The term "fresh" is a term of art in cryptography.  It means, roughly,
>>>>> "not used previously."
>>>>
>>>> So this implies that client must keep track of already used keys? it
>>>> imposes limit on length of connection?
>>>> True it or false, It should be explicitly clarified in the text.
>>>
>>> The normal practice in cryptography is to just use large enough values
>>> such that the probably of collision is sufficiently small as to be
>>> acceptable.  For example, if you use a 20 byte nonce, the probably of
>>> collision is zero for all practical purposes.
>>>
>>> This stuff is all extremely normal.
>>>
>>> Adam
>>>
>>
>