process(_at_)qz(_dot_)little-neck(_dot_)ny(_dot_)us (Eli the Bearded) writes:
Philip Guenther <guenther(_at_)gac(_dot_)edu> wrote in another thread:
[question about how to use +box addressing in procmailrc files]
It shows up as $1 in your procmail, which you can then test with
something like:
ARG = $1
:0
* ARG ?? ^^pgp^^
pgp-folder
Privately I asked where this was documented and he pointed me towards
a paragraph in one of the man pages that "implies" this. I see no such
implication there, but maybe I am not reading between the lines enough.
The semi-logic of the implication runs along the lines of:
"var ??" requires a variable name on the left hand side. Meanwhile,
$=, $_, $@, $1, $2, etc are listed as _expansions_, with no mention
of variabled '=', '_', '@', '1', etc.
Yeah, it's kinda weak. If you can come up with a better description
for the procmailrc(5) manpage, send it to Stephen. Be aware, that
Stephen is pretty strict about what he'll accept. I would say that
he believes in not stating anything that is deducible or that "would
be the obvious way of doing it". I could see him accepting a suggestion
that a period be inserted into the first sentence of the first paragraph
of the BUGS section between "$\name" and "$#," (and then corrected
grammatically) so as to indicate that $# and following are not environment
variable expansions, as the previous items in the list are.
In my tests this does work, although I had to remove the carets. And
The carets make the condition match iff ARG is _exactly_ "pgp". If
you want it to match anything that contains "pgp" then you wouldn't
them.
it even works for Bcc'ed stuff which has no indication of the envelope
address in the headers. What else works? Do $0, $2, $3 etc have useful
For normal local delivery, $1 is the only one that contains anything
interesting, being just the "+detail" part of "user+detail" addresses.
If procmail is invoked via the procmail 'mailer' from sendmail, the $1
contains the envelope sender, and $2, $3,... contain the full envelope
recipient addresses (at least one, possibly many).
stuff? What is required for procmail to know $1? I doubt it is going
Procmail must be invoked as the local mailer from the sendmail.cf,
with a newish sendmail binary and config. The sendmail RELEASE_NOTES
file indicates this was added in 8.7. Remember: the _config_ file
is where most of the support for this is, not the sendmail binary.
to be there for Bcc'ed stuff when procmail is called from a .forward
file. And I see no way it could be there when procmail is called to
process an existing box of Bcc'ed messages. Why is this not
documented? What else isn't documented?
Because no one has written the documentation. What else isn't documented?
Uhhh, everything? That question has an infinitely long answer, and it's
not clear which part you're interested in.
please do not CC me when replying to the list
How about if I To: you and Cc: the list? (Just kidding)
Philip Guenther