[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20130127023513.GA3905@openwall.com>
Date: Sun, 27 Jan 2013 06:35:13 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: formats that duplicate dynamics
Dhiru, Jim, magnum -
The "postgre" format at the very least needs to be renamed from the
ridiculous "postgre" to "postgres". This includes the filename,
FORMAT_LABEL, and the "$postgre$" prefix string.
However, there's a more general issue: can't the same be implemented as
a dynamic format? In fact, doesn't one of the predefined dynamic formats
support these same hashes already? Maybe we should just document what
predefined dynamic format should be used, and make this prominent in its
FORMAT_NAME (so that it'll be seen on --test runs)?
This general issue probably applies to several other formats as well,
including e.g. "mongodb". And we know for sure it applies to "radmin" -
we have both this format and the corresponding dynamic, clearly
documented as such.
Dynamic is faster per-core: it uses SIMD, whereas Dhiru's formats
don't. However, it currently does not use OpenMP, whereas Dhiru's
formats do. Is this reason enough to keep both for now, as separate
formats? Even if so, shouldn't we have the separate formats that
essentially duplicate dynamics use the corresponding $dynamic_*$ prefixes
instead of their own ones? This brings up the issue of which of the two
formats should be the default, though - but maybe that's a better issue
to have than not even informing the user of the possibility to use a
dynamic for the same thing (and better per-core speed), like what we
have now.
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.
