[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20120610134800.DOK4B.25538.imail@eastrmwml213>
Date: Sun, 10 Jun 2012 13:48:00 -0400
From: <jfoug@....net>
To: john-dev@...ts.openwall.com
Cc: Frank Dittrich <frank_dittrich@...mail.com>
Subject: Re: ./john
--markov=399:9876543210987654321:12345678901234567890:30 error
---- Frank Dittrich wrote:
> While working on my --markov=section version, I noticed a problem which
> exists in the --markov=level:... version:
>
> $ ./john --markov=399:9876543210987654321:12345678901234567890:30 --stdout
> Warning: End = 9223372036854775807 is too large (max = 165139712673862902)
> Error: MKV start > end (9223372036854775807 > 165139712673862902)
>
>
> I have the same problem when someone tries to specify such large values
> for MkvStart or MkvEnd in the [Markov:section]
>
> Do we need to fix this? I yes, what would be the best fix?
I have never really liked the 'huge' numbers in markov myself, but do know why simon wrote it that way.
I had always thought it would be better to have some percentage. So it would be like this:
--markov=399:0:10:30
--markov=399:10:20:30
...
--markov=399:90:100:30
This would split up the first 10% of the candidates into the first run, and so on. The nice thing about this, is no matter how many candidates there are totally to do, the percentage does what is needed. No need to precompute the counts, and do the division by hand.
If we had percentage start-stop, then
--markov=299:0:10:30
--markov=299:10:20:30
...
--markov=299:90:100:30
would work just a good as the runs which were at 399 level.
I have thought about doing something like this in markov, but have not ventured any further than thinking about it.
JIm.
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.
