problem with prefs

David Shaw dshaw at jabberwocky.com
Tue Feb 11 14:24:01 CET 2003


On Tue, Feb 11, 2003 at 01:47:47PM +0100, Janusz A. Urbanowicz wrote:
> On Tue, Feb 11, 2003 at 07:10:46AM -0500, David Shaw wrote:
> > On Mon, Feb 10, 2003 at 11:28:15PM +0100, Janusz A . Urbanowicz wrote:
> > > alex at FUCKUP:[~]:4:0:> gpg -v --version 
> > > gpg (GnuPG) 1.2.0
> > > []
> > > Hash: MD5 (H1), SHA1 (H2), RIPEMD160 (H3)
> > > []
> > > 
> > > ~/.gnupg/gpg.conf contains the following line:
> > > 
> > > personal-digest-preferences h3
> > > 
> > > but when I do gpg --clearsign testmessage:
> > 
> > [..]
> > 
> > > It is still hashed using SHA1 instead of RIPEMD-160 I selected. Why?
> > 
> > The personal-xxxxx-preferences commands only apply during operations
> > that involve a recipient.  Otherwise, --digest-algo is used.
> > 
> > However, read the gnupg-users thread entitled "Personal prefs" from
> > November of 2002.  Someone asked the same question back then, and the
> > development GnuPG was changed to consult personal-digest-preferences
> > if digest-algo was not set.
> 
> yes, I remember the thread, especially what you wrote:
> 
> | Yes.  Putting cipher-algo or digest-algo in gpg.conf should really
> | only be used in very special circumstances.  The prefs functions let
> | you do the same thing, but in a safe manner that won't generate
> | messages that violate the recipients preferences.
> 
> So, there's not a good way to use RIPEMD160 for clearsigning without risking
> a major breakage?

Not in the stable branch.  1.3.x lets you use
personal-digest-preferences in the way you want.

That said, RIPEMD160 happens to be supported fairly widely (i.e. in
PGP), so it would not be a major breakage to use it, though it might
be a minor one.

> I know it is a Good Thing to emit messages that have undefined recipient to
> have minimal protocol requirements for the recipient, but in the case I'd
> like to switch the hash. Why does perconal-digest-preferences not affect
> clearsigning? A deliberate design decision? I'm curious.

More of an evolution of features thing than a deliberate design
decision.  The personal-xxxx features grew out of a desire to sort the
preferences differently and not as an alternate way to specify
algorithms.

David

-- 
   David Shaw  |  dshaw at jabberwocky.com  |  WWW http://www.jabberwocky.com/
+---------------------------------------------------------------------------+
   "There are two major products that come out of Berkeley: LSD and UNIX.
      We don't believe this to be a coincidence." - Jeremy S. Anderson




More information about the Gnupg-devel mailing list