[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kmail-devel
Subject:    Re: [Aegypten] Bogus messages about certificates?
From:       Ingo =?iso-8859-15?q?Kl=F6cker?= <kloecker () kde ! org>
Date:       2002-09-30 19:10:33
[Download RAW message or body]

On Monday 30 September 2002 09:06, Werner Koch wrote to 
aegypten-intern@intevation.de:
> On Sat, 28 Sep 2002 14:07:01 +0200, Bernhard Reiter said:
> > Needs a good answer.
> > Some things are BUGS (some already known for a long time).
> > Some crypto things about SPHINX need to be explained.
> >
> > From: Ingo Klöcker <kloecker@kde.org>
> > Subject: [Aegypten] Bogus messages about certificates?
> > To: kmail@kde.org, gpa-dev@gnupg.org
> >
> > But the signature on a message which was signed with a certificate
> > that
> > had not expired by the time the message was signed will always be
> > valid
> > (as long as the certificate isn't revoked).
>
> No.  The signature is not valid after the expiration time of the
> certificate.  The reason to use an expiration time is to make it
> impossible for an attacker to issue a signature using a compromised
> key belonging to an expired certificate.  The way a signature
> verification should be handled in the workflow of an office is by
> registering valid signed documents by other means.

I understand. This would mean that every incoming document would have to 
be signed with a local key which of course must never expire. Do you 
know of any MUA or MTA that does this?

> The warning we display should mention that the certificate has been
> valid at the *claimed* time the signature has been created but
> nevertheless we can't know anything more. I think we agreed on
> displaying it in green with that warning.

That's fine with me. But then we might also add a warning that a valid 
signature which was made with a not-expired key could also be forged. 
Nothing is 100% secure. I don't think it's more likely that an expired 
key is compromised than that a not-expired key is compromised. In both 
cases the key must be revoked immediately. In my understanding 
"expired" simply means "isn't used anymore". It does not mean "could 
have been compromised in the meantime".

> > Is this also a Sphinx requirement that the certificate has to
> > contain the email address of the sender?
>
> Yes.

Then you should probably implement the function which checks this. ;-)

> > To check whether an email message really came from the sender one
> > simply checks the signature. The address in the From: header is
> > completely irrelevant. Of course, it might be worthwhile to warn
> > the recipient
>
> No it isn't.  A MUA uses the From/Reply-To header for an reply.  This
> can be used to trick the recipient into replying to the wrong person
> and for example agreeing on a contract.

Then the receiving MUA should also complain if the From/Reply-To header 
doesn't match the key. I don't see where you implemented this in KMail. 
;-)

> > when the sender's address isn't contained in the signing
> > certificate. But this certainly doesn't make the signature invalid.
>
> No it is not invalid, but a note should be printed.

O.k., but with the option to "Don't show this warning again.". At least 
for not paranoid KDE users.

> > BTW, do we really have to call it certificate? In the OpenPGP world
> > this is usually called key instead of certificate.
>
> Agreed, however certificate is more exact and so I don't see an
> urgent reason to change it.

Well, the reason is that we use "key" everywhere else in KMail and that 
"certificate" isn't common language while "key" is. Even GnuPG uses the 
term certificate only for revocation certificates.

But today the message freeze began. So we shouldn't upset the 
translators and leave it for now.

Regards,
Ingo

[Attachment #3 (application/pgp-signature)]
_______________________________________________
KMail Developers mailing list
kmail@mail.kde.org
http://mail.kde.org/mailman/listinfo/kmail

[prev in list] [next in list] [prev in thread] [next in thread]