ietf-openpgp
[Top] [All Lists]

Re: "Yes, I can handle PGP/MIME"

2004-04-16 08:10:45

Adrian 'Dagurashibanipal' von Bidder wrote:

Why can't you invent a convention that adjusts the
userid to include an OpenPGP/MIME hint?


Ugh. And then I'm forced to change the userid when I change my mailer, and lose all signatures on the userid that I've collected so far? I think embedding this info in the selfsig is the way to go - I can replace that and if the software is clever enough to silently ignore all but the newest selfsig, all users won't even notice that the underlying technology doesn't allow you to change your key, but only to add to it. I still think that this is a perfect example where notations would be the solution.


Well, right.  All that above has to be balanced
against the fact that email is a user application,
and it's not good to pollute OpenPGP with special
hacks and bits.  (I'm not wedded to my above
suggestion, it's more in the vein of searching
for alternates.  And there seem to be plenty of
bits available for this sort of use...)

If the choice were between adding a bit as per the
original thread suggestion by David, and overloading
another bit already utilised, as a "version" indicator,
(the "preferred keyserver attribute" ?) then I'd
definately plumb for the former - define a special
bit:

  0x02 - Recipient is capable of handling OpenPGP/MIME (RFC-3156).

(etc.)  I think David's original post still rules.

As a standard, it's good to keep an eye on
compatibility amongst implementations, but
overloading should be discouraged.  There's
no reason why future implementations couldn't
adopt that bit, even if they used some
overloaded bit in the past/absence.

iang

PS:  at what point do we go for feature freeze?
How long does this process of minor additions
go on for?  Derik, what is the process to get
this thing signed off and passed into law?


<Prev in Thread] Current Thread [Next in Thread>