[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20130926023018.GA22890@openwall.com>
Date: Thu, 26 Sep 2013 06:30:18 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: External node distribution (call for help)
magnum,
On Thu, Sep 26, 2013 at 04:14:19AM +0200, magnum wrote:
> Not sure what approach you mean. The my_words & their_words are an
> approach from wordlist.c but in that case we have a proper line number
> too (much like seq) and I hoped to avoid incrementing that. Hmm but I
> see this now:
>
> /* Restored session. line_number is right after a word we've actually
> used. */
> ...
> /* We assume that line_number is at the beginning of other nodes' block */
>
> Does this apply to external mode too?
With a sufficiently similar approach and implementation, it would.
> I mean, can I assume that I am at
> the start of a full "their_words" block after resuming?
In an equivalent of the "if" branch that you extracted the above comment
from, yes.
> How/why is that?
> I can't see why that would be a safe assumption. A fix_state() seems to
> happen right before a clear_keys and if "we" are several nodes (as in
> -node:1-7/10) that could be in the middle of my_words.
The "We assume ..." comment is located inside an "if". The special case
that you mention here is the "else" after that "if". That case is
handled as well, and you may handle it in a similar fashion in external
mode.
Alexander
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.
