[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 30 Apr 2011 03:53:40 +0200
From: magnum <rawsmooth@...dband.net>
To: john-users@...ts.openwall.com
Subject: Re: JtR 1.7.7 & 1.7.7-jumbo-1 + MPI error compilation
OS X
On 2011-04-29 06:55, Solar Designer wrote:
On Thu, Apr 28, 2011 at 08:57:35AM +0200, websiteaccess@...il.com wrote:
> > I have downloaded:
> >
> > john-1.7.7.tar
> > john-1.7.6-fullmpi8-after-jumbo3.diff
> > john-1.7.7-jumbo-1.diff
>
> As usual, no one has updated the MPI patch to the new version of JtR
> yet, and I don't know whether anyone intends to.
W.A - 1.7.7 as of today does not add any new formats, performance or
major functionality. On the contrary, you lose intrinsics and enhanced
Markov. Just continue using 1.7.6 for now.
I do intend to post an update of fullmpi, but it's low prio. I would
prefer to see a Jumbo with stuff like intrinsics, new Markov and utf-8
added (not to mention Jim's proposed core changes) before posting an mpi
patch because I'm not very keen on "maintaining" public versions of mpi.
Not that I have any private things in my tree, it's just that it's a
fairly long (and boring) procedure to rebase it to plain jumbo and
regression test that.
> All - I don't mind merging the MPI patch into jumbo, which will
> apparently make life easier for W.A. Thus, please feel free to make and
> submit a revision of the MPI patch suitable for integration into the
> jumbo patch. It must not strictly depend on MPI; instead, it should
> have -DHAVE_MPI, which could be uncommented in the Makefile to activate
> the functionality (similarly to HAVE_SKEY and -lskey, which we already
> have - these came from another patch I merged).
Actually I alomst finished an #ifdef'd fullmpi patch for 1.7.6 but I
felt it was too intrusive to include in Jumbo so I never posted it. But
I do use mpi regularly for non-OMP formats. Maybe I should reconsider.
It would be a piece of cake to Jumbo-ize the old "mpi10" patch
(incremental only) but that one has little real value. The fullmpi patch
is rock stable and scales near perfectly with several modes but it does
make a mess with the sources, I'm not sure you would really merge it if
you had a closer look :) OTOH as long as it's ifdef'ed, why not...
I'll have another look at it.
magnum
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
