To get off trust models and e-mail addresses for a moment ...
Is there a smart PEM-ish way to handle Bcc? That is, sending a copy of a
message to another recipient without the main recipients knowing about it.
If a single encrypted message is sent to a number of recipients, then they
all need to be listed in the PEM message body, which would disclose the
identity of Bcc recipients to the main recipients. Even if only an
encrypted key field (no name) was included, the main recipients would
know that someone else has seen it, and depending on the message they
could then determine who.
Would the correct course of action be to trap Bcc in the UA and send
separate messages, one to the main recipients, and one to each of the Bcc
recipients? One drawback is that the copies to the Bcc recipients would
not have the addresses of the main recipients in the header. So, should
the Bcc message be a forwarded MIME message/rfc822 copy of the main message,
or something else?
What would be "nifty" would be a way to send the message once to all of
the recipients with just the main recipients listed in the headers and the
message body, and then send along a separate message containing the
encrypted key to each of the Bcc recipients so they can decode the main
message. Too radical? Probably.
Inquiring UA design team minds want to know.
Cheers,
Rhys.