Rereading this thread, I think I was not apologetic enough for my original incorrect reading of what the "RIPEMD-160" standard was about.
(My original, superficial, understanding was that this was a family of hashes of 160 bit hashes. But I was wrong about that. And I have not studied the standard well enough to know whether the structure of an sha1 implementation is of any use in a ripemd-160 implementation.) Anyways, I was wrong, and you (Pascal) corrected me, and I should have been more appreciative of that. I really do like the corrections, I'm just slow to recognize that sometimes. Thanks, -- Raul On Thu, Jun 26, 2014 at 9:51 AM, Raul Miller <[email protected]> wrote: > Yes, it's a different hash, but it's related so perhaps of interest. > > Also, future implementations of J might offer significant speed > improvements. > > That said, if speed is truly a concern, you should probably be using > dedicated hardware (e.g. something built on FPGAs) and not a general > purpose computer. > > Thanks, > > -- > Raul > > On Thu, Jun 26, 2014 at 9:46 AM, 'Pascal Jasmin' via Programming > <[email protected]> wrote: > > FYI, RIPEMD is a different function than SHA1. Also, an openssl binding > is much faster, and cross platform. The required dll is also now bundled > with J802 on windows, and iirc, is dependably available on linux and osx, > and maybe android. > > > > > > ----- Original Message ----- > > From: Raul Miller <[email protected]> > > To: Programming forum <[email protected]> > > Cc: > > Sent: Thursday, June 26, 2014 9:35:12 AM > > Subject: Re: [Jprogramming] RIPEMD-160 hashes > > > > Actually, there's a bug in that implementation: > > > > It should be something like: > > > > sha1=: [: process/ H (,~ |.) _512[\] > > > > And, 'process' should be made into a dyad which takes H as y and the > > current block as x. > > > > In other words, the current implementation computes independent sha1 > > sums for each block, instead of computing a single sum for the > > sequence of blocks. > > > > Does anyone feel like registering for rosettacode and fixing the page? > > Oh, and testing the code to make sure my thinking isn't broken, again? > > (If not, I'll get to it sooner or later.) > > > > Thanks, > > > > -- > > Raul > > > > > > > > > > > > On Wed, Jun 25, 2014 at 3:20 PM, Raul Miller <[email protected]> > wrote: > >> http://rosettacode.org/wiki/SHA-1#J seems to qualify as an RIPEMD-160 > >> hash algorithm. > >> > >> Thanks, > >> > >> -- > >> Raul > >> > >> > >> On Wed, Jun 25, 2014 at 11:54 AM, John Baker <[email protected]> > wrote: > >>> I am wondering if anyone has invoked RIPEMD-160 hash algorithms from J. > >>> > >>> There is public C source for this hash > >>> > >>> > http://cpansearch.perl.org/src/CHGEUER/RIPEMD160-0.02/rmd160/doc/ripemd160.html > >>> > >>> and looking through it is enough to convince me that this is best left > in > >>> C. It seems most of the implementations are tailored for 32 bit > machines > >>> and I need to run this from a 64 bit machine. Any suggestions will be > >>> welcome. > >>> > >>> > >>> -- > >>> John D. Baker > >>> [email protected] > >>> ---------------------------------------------------------------------- > >>> For information about J forums see http://www.jsoftware.com/forums.htm > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
