>> On Aug 20, 2017, at 8:35 AM, Robert Moskowitz <[hidden email]> wrote:
>> It is 64 - 160 BITS
> Correct, with the word "cryptographically random" somewhere in
> there, for at least 64 of the bits.
>> Which is 8 - 20 OCTETS
> Correct, since an "octet" is 8 bits.
>> or 4 - 10 BYTES
> No, a "byte" nowdays is the same as an "octet", though there have been
> variant definitions of byte, while "octets" have always been 8 bits.
I am going back to bed.... :)
But my bit collision analysis still holds true. Collisions are not a
concern if openssl rand is a good prf.
Re: Clearing up some of my mistakes on serial number
In reply to this post by OpenSSL - User mailing list
On 08/20/2017 09:50 AM, Salz, Rich via openssl-users wrote:
> If you generate 19 bytes or RAND output, it will never exceed 20 bytes encoded.
> OpenSSL will be generating 159 bits of RAND output, so that it will never exceed 20 bytes encoded. The command-line RAND program is bytes, the C API is bits.
So Viktor reminded me about bits, bytes and octets.
Too much on the brain. No real excuse.
Back to the drawing broad.
8 bits is a byte. 8 bits is a byte. 8 bits is a byte. 8 bits is a byte.
Two hex positions is a byte. One hex position is a nibble.