Re: The issue of mime parsing

Rickard Armiento <[email protected]>:

>> this is tempting as the delivery of message parts can be
>> de-coupled, but the sender may already have specified details of
>> the encryption. but we need both:  senders must be able to do this,
>> and receivers must be able to re-negotiate.
>
> I do not understand. Why do the sender need to specify details of
> the encryption on beforehand? Why can't the sender just be satisfied
> knowing that each time a receiver connects to his server to fetch
> data, it will negotiate a suitable encryption for that session?

that, of course, would be best, but when the sender wants encryption
to the receivers public key, there aren't any choices left to
negotiate.  im2000 should not confine users to one or two methods:
this has to do with the trust-model a particular user wants to use
(or exclude!) and it should also allow new methods to be introduced.
some models are very strict, like symmetric encryption using
previously exchanged passwords.

but maybe you think the sender limits possible choices for the
receiver to pick from?  yes, this is ok.

  clemens