[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-id: <5EF144C4-0A61-4F59-8888-0EDA1245FEDB@mac.com>
Date: Sun, 06 Apr 2014 15:22:53 -0700
From: Larry Bugbee <bugbee@....com>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] Proposed changes to author's code
I just started. At https://github.com/buggywhip/yarn_py you will find the following.
yarn_wrapper.py is a ctypes Python wrapper for the Yarn Password Hashing Candidate.
yarn_demo.py is a simple demo program to illustrate various calls.
yarn_kats.py converts and runs the provided KATs.
To compile an OSX .dylib:
gcc -dynamiclib -arch x86_64 -o libyarn.dylib yarn.c
Python wrapper API:
from yarn_wrapper import *
hash = yarn(outlen, pswd, salt, pers,
t_cost, m_cost, par, initrnd, m_step)
# Note: this version of blake2b does NOT support a key
hash = yarn_blake2b(outlen, pswd, salt, pers)
The goal is to now create a Python wrapper to support the generic API.
--------------------------------------------------------------------
On Apr 6, 2014, at 12:21 PM, Bill Cox <waywardgeek@...il.com> wrote:
> I have compiled each PHS with a common main.c. Doing so required
> minor changes to the following PHS's:
>
> Argon
> Battcrypt
> Catfish
> Gambit
> Lanarea
> M3lcrypt
> MCS_PHS
> OmegaCrypt
> Parallel
> PolyPassHash
> Pufferfish
> RIG
>
> My goal is to have each entry have a PHS function that is C linkable.
> This is required, for example, if we wish to have bindings in
> different languages like Python, so I made the minimum changes to make
> these C linkable. The C++ projects mostly just needed extern "C" in
> their PHS declarations. Some of the others had minor portability
> issues that I fixed, possibly not exactly how the author would prefer.
> My change to Yarn were magically identical to the V1 version of the
> code... spooky!
>
> I've attached a diffs file, which shows all the changes I made to each
> project. I also added the V1 versions of Yard, M3lcrypt, and POMELO,
> so if my changes get pulled upstream, these will also get pulled. I
> think there is a ^M line ending issue in one of them that I haven't
> fixed.
>
> If the authors could review the diffs and let me know if they are OK
> with the mods, I'll ask that these changes get pulled into the main
> git repo.
>
> Thanks,
> Bill
> <diffs>
Content of type "text/html" skipped
Powered by blists - more mailing lists