At least with most IMAP servers, the local storage mechanism is either
mbox, MH, or Maildir format. Procmail (with patches, for Maildir) can
support all three.
Still don't see the problem...
Let's say I have an IMAP server running on a separate, user-inaccessible
machine. I would like people to be able to use procmail filters to have
their mail delivered into different IMAP mailboxes. The most elegant way
to do this, it seems to me, is to use IMAP's "deliver" agent, which already
exists, is integrated into IMAP and does everything right. So let's say
I build a little program that runs on the user-accessible machines and
takes as input a mail message and connects to IMAP's delivery agent in LMTP
mode and tells it to deliver the message to a given mailbox. Yes? Then
the users' procmail filters can invoke this program and off goes their mail.
BUT. I encounter two problems. One, my users now have to explicitly invoke
this program in every rule. In addition to the fact that this means that
when I switch to IMAP everybody has to re-write their filters, this is very
error-prone. It would be much nicer if old simple rules continued to work,
but folder names simply were, for now on, interpreted as names of IMAP
mailboxes. Two: let's say the user gets a message that is of a kind that
IMAP rejects. Procmail passes it to my little program, which passes it to
IMAP's delivery agent, which rejects it. My little program generates
a sendmail-compatible error exit code. How do the users write their filters
so as to make procmail pass this exit code back to sendmail (on every rule)?
Is there a natural way to do this?
-malgosia