At 3:11 PM -0800 1/26/99, Ned Freed wrote:
What's wrong with saying that some actions affect delivery status,
and other actions don't? Actions that affect delivery status are:
keep, discard, forward, reject. Actions that don't are: reply, mark
(if this extension is supported).
The issue isn't so much with the present collection of actions (although
I suppose we could go back and forth quite a bit about whether or
not a reply implies reply and discard), but with future ones. Do we
really want to allow each action someone adds to say whether or not
it affects delivery status?
If so, I guess I can live with it. But I'm dubious as to whether or
not the complexity (remember, you're opposed to complexity ;-) is warranted.
I agree the issue is really with extensions.
My point is that we need some consistent basis for determining the
default delivery status. Even if we choose to ignore it, we really
do have two classes of actions: those that explicitly affect delivery
(that is, cause it or prevent it), and those that do not. If we
really want to say that delivery happens by default only if no action
whatsoever is specified, what we are saying is that all actions which
do not explicitly cause delivery implicitly cause non-delivery. If
this is really what we mean, than we need to state it. Otherwise it
is ambiguous and implementation-dependent.
Personally, I think it would be counterintuitive and dangerous to
make this the position, but I also feel that not being clear is much
worse.
I think a clear and safe way of phrasing this is: Message are
delivered unless an action is performed which explicitly prevents
delivery (such as "reject").
We can separately talk about the default delivery mailbox: The
default mailbox for delivery is "INBOX" on an IMAP server, the
maildrop on a POP server. If FileInto is supported, this can be used
to alter the delivery mailbox.