(Warning: 4:30Am rambling ahead ;)
On 15 Jul 1997 21:39:58 -0000, you said:
What is the desired behavior if person A posts to a mailing list, and person
B follows up, etc etc - and then person G replies to F, trims out persons
A through E, and does a *BCC:* to the mailing list?
People do this as a crude way to control followups. Given widespread
support for a followup field, why would someone do such a Bcc?
That's not the question I meant to ask. What I *meant* to ask was:
Assuming even 90% of the MUAs are updated to support a followup field,
how robust will it be if somebody in the 10% uses the bcc chainsaw
I see any algorithm that basically merges the from/to/cc's (with
poissibly additional local knowledge) as being very crippled if a bcc
attack occurs. On the other hand, a Followup: header will be a bit
more robust - if the user's MUA doesn't support followup:, the header
will likely not even be displayed by default, so naive users can nuke
and bcc: to heart's content (at least in regards to "forward"ing a
piece of mail.
How robust does it *need* to be? Is "95% effective duplicate suppression"
good enough? Or do we want to be hard-line, and design it so people can
get the followup correct even after significant user mangling of headers?
A few more random thoughts:
(a) has anybody considered reply/forward and the difference between
them for followups? Is the usual "Just add soem Resent- tags" scheme
the Right Way to do this, or does forwarding require fixing up the Followup:?
(b) what are the semantics of followups for an email that ends up
encapsulated as a message/rfc822 in a digest or similar? Note that the
digest itself may have no followups, but you can then proceed to cite
more than one posting (possibly with divergent followups). What about a
message/partial if the parts have disagreeing Followup:s?
(c) we also need to think about if we *do* create a new Followup:
header, what its interactions with bcc: and disclosing the recipient
list really are (yes, I know different MTAs handle concealing bcc:
differenty - this just has the potential of making it even worse).
(Why would someobdy want to do that? I can easily see somebody sending
out something "official", and bcc:ing his boss, and adding the boss to
the Followup: so that the boss can lurk in the subsequent discussion.
Bad move if the boss wanted to *really* lurk. Yes, better handled in
other ways. Yes, users will do it anyhow ;)
/Valdis (who is just awake enough to think of boundary conditions
that he can't solve ;)