Hi Zooko,

> Now I've learned that this is not true,
> neither for RandomPool nor for X917RNG.
RandomPool is based on PGP's Random Pool.

> ... use X917RNG by passing a correctly-sized array of
> zeroes for the "deterministic time vector" argument
If I recall correctly, ANSI 9.17 recognizes two sources: 3DES (E-D-E)
and SHA1. In either case, I think the time variant is required.

> after a brief search I didn't see a block
> cipher with a sufficiently large (256-bit)
SHACAL-2 is 32 bytes, but I'm not sure how secure it is. I don't use
it so I have not read up on it.

Jeff

On 3/13/08, zooko <[EMAIL PROTECTED]> wrote:
>
> Folks:
>
> I've continued to use and extend my "pycryptopp" library which
> provides a small subset of Crypto++ functionality to Python programmers.
>
> I've also set up a "trac" to manage pycryptopp.  Trac is a
> combination of an issue tracker, revision control history, source
> code browser, and wiki:
>
> http://allmydata.org/trac/pycryptopp
>
> Notice the "Buildbot" button -- that takes you to the display of the
> results from automated unit tests.
>
> Recently I realized that my code which was intended to offer the user
> to deterministically generate a public/private key-pair from a secret
> seed was not working.  I had assumed from the names that "Auto
> Seeded" RNGs drew from information outside of the explicit arguments
> and that therefore non "Auto Seeded" RNGs would be determined
> entirely by their arguments.  Now I've learned that this is not true,
> neither for RandomPool nor for X917RNG.  (So it would be good to
> augment the documentation to make it clear that the RandomPool and
> X917RNG both sample the clock by default.)
>
>
> So I started to use X917RNG by passing a correctly-sized array of
> zeroes for the "deterministic time vector" argument, but then I
> realized that the seed size is limited by the block size of the
> BlockTransformation, and after a brief search I didn't see a block
> cipher with a sufficiently large (256-bit) block size that I wanted
> to use.  (It would be nice if there were a convenient way to specify
> "deterministic PRNG only, please" without having to construct a 32-
> byte array of zeroes.)
>
>
> I tried using SHA256, and then Salsa20, but of course they aren't
> BlockTransformations.
>
> So my next idea is to define my own BlockTransformation class which
> uses Salsa20 and pretends to have a block size of 32.
>
>
> Make sense?
>
> Thanks!
>
> Regards,
>
> Zooko
>
> >
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the "Crypto++ Users" 
Google Group.
To unsubscribe, send an email to [EMAIL PROTECTED]
More information about Crypto++ and this group is available at 
http://www.cryptopp.com.
-~----------~----~----~----~------~----~------~--~---

Reply via email to