Ridge Cook wrote:
It could also be overcome by adding a button on
thunderbird to auto-generate the cert and install
it....
Well, you are absolutely right. A self signed cert mechanism could easily
be included in a browser, mail client...or PGP for that matter.
that they aren't could be multiple factors...and I just don't have the
patience to parse out the inside politics..
But for the life of me, I don't understand PGP/GPG's reluctance to do this.
I also don't like the knee jerk reaction to a more formalized PKI structure
evident in the PGP community. I've argued the point over and over and you
end up hitting a brick wall that boils down to-
CA (public or private)=greedy corporate money grubbers=bad
Its like their brain gets stuck on the term. Makes no sense.
The OpenPGP community is .. yes, if anything
even more religious than the x.509 community.
I think they have some cause; their original
development was highly influenced by validated
threats to human rights workers. If you talk
to some of the insiders in that community,
they have tales of people who've died and/or
disappeared primarily due to poor commsec,
and believable scenarios where MITM might
occur.
But, making the jump from "people who could
die" to "the rest of us" is still a hugely
important and necessary step which they simply
fail to make. I'd never thought of it like
that, but, yes, a GUI-based OpenPGP client
should also bootstrap up and generate an
initial key ready for action. Just by the
nature using a GUI, we've already accepted
that a certain amount of security is exchanged
for convenience.
I think the reason that you run into brick
walls there is that security models are way
too involved, complex and subtle. Once they
are laid down, concrete is applied and walls
are built. I also have been bashing my head
against these things and people really get
nervous when their assumptions are shown to
be wrong (e.g., the CA model in S/MIME is
based on a no-prior-relationship assumption
from HTTPS, which is suspicious in HTTP,
but totally wrong with email).
However , as you say, just for email a Root>User format is un-necessary.
Look at the limited key signing in PGP...Most email does not matter that
much. But since x509 is THE networking standard; formality is required to
allow confident network access or for commercial activity.. Does it have its
problems and issues? Yes, but so does any other authentication protocol.
I'm not sure where you get the notion that x509
is "THE networking standard." I'd certainly like
to see some stats on its use that support that.
I grant you it is the popularly believed one,
but actual usage is not matched by popular belief.
For example, in the HTTP world something around
1% of servers deploy certificates, so it doesn't
classify as a successful standard in my book.
S/MIME almost certainly is pipped by OpenPGP.
In the secure remote terminals world, secure
telnet is as dead as a dodo, ordinary telnet
is sometimes used but in practice everyone uses
SSH. And, in the developing p2p world, it's all
open.
However, for issue presented in this thread, xCA is a great utility, offers
several hash choices in key formation , unlimited key size, nice gui, 3des
protected database...... . Just the thing Ms. Knoell needs if she does not
want to go through the hassle of setting up a certificate server.
Agreed.
iang
_______________________________________________
mozilla-crypto mailing list
[EMAIL PROTECTED]
http://mail.mozilla.org/listinfo/mozilla-crypto
