On Sun, 10 Sep 2000, Michael J. Tubby wrote:
> if personal alias mike@??? alias mike@??? alias
> mike.tubby@??? alias mike.tubby@???
> then
> save personal/
> finish
> endif
That will save to a directory called $home/personal/ which is not what
you want, as I understand it.
When an appendfile driver is called from a .forward, the entire path
name is what the .forward file supplied, and any configuration that you
may set in the transport's "file" or "directory" options is ignored.
> Perhaps I have mis-understood the interaction between the directors and
> transports
> for this?
Yes, that's it.
> In the .forward file I have a
> 'save' directive
> with a directory name which ends in a slash. This should be recognised as a
> directory name.
In the .forward file, this triggers the address_directory transport
instead of address_file. Within the transport, the terminating /
triggers the directory code instead of the file code. However, the path
is what came from the .forward file.
> 1. does the directory name in the 'save [dirname]/' end up back in the
> $address_file
> variable as I am expecting?
It should, but the transport also just uses the value directly, without
looking at "file" or "directory".
Hmm. If I were starting again, I'd do it your way. Originally,
$address_file did not exist, you see. It got added quite a lot later,
for some reason or other.
It's late to change now, but one possible, incompatible change would be
to say that if file/directory were set, they should override. (The
default transport has no setting of either of these.) I will put this on
the Wish List.
> 2. if not, then how do I communicate the name of the directory back to the
> transport
> which has to deliver it?
At present, you must give more of the path.
> 3. where is my mail going when it vanishes?
Into your home directory, I suspect (without the Maildir subdirectory).
> Any help with correcting my configuration appreciated...
You might have got further understanding what was going on if you had
run a delivery with -d9, to get the debugging output. It would have
contained the file name somewhere.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
