procmail
[Top] [All Lists]

Re: LASTFOLDER blues

2007-03-26 10:54:09
On Mon, Mar 26, 2007 at 06:40:53PM +0200, Dallman Ross wrote:
What use is this for a bad fs, an fs over quota, or a full fs? The
issues are not someone having wrongly set owner and file modes,
although you can add this under the category user-errors-that-
shouldn't-cause-mail-to-be-lost.

Maybe.  But that, too, is an arbitrary design choice.  You
want to requeue messages for an over-quota account.  The mail
will loop through your system indefinitely congesting your
server bandwidth and job queue.

No, I want it to TEMPFAIL until the sender gives up and bounces. 

What is so evil about asking you, the sysadmin, to code in
explicitly what you want procmail to do?  Run your procmail.
If you need to conduct special fs checks at the end -- *before
you purposefully throw the message away that procmail has not
yet abandoned!!* -- why, do so!

Sorry, no. I'm not going to workaround a problem by doubling my 2000
line recipes. This is a problem that clearly belongs into the MDA's
domain and any "workaorunds" like the pairing is error-prone and a
hack - well, not even deserved the name "hack" IMHO.

As Bart said, even if there is something to your argument, you
have three choices, only one of which seems viable for you right
now: (1) put a kludge in place to test for your error condition
and deal with it; (2) patch procmail (if you write a nice patch,
be sure to share it with us); or (3) get Stephen or Philip,
or somebody else, to un-mothball the procmail project and work
this out to your satisfaction.

I'll go with (4): Admit that unfortunately procmail has long reached
and outlived its EOL and learn something new.
-- 
Axel.Thimm at ATrpms.net

Attachment: pgpgZzOFAMGFY.pgp
Description: PGP signature

____________________________________________________________
procmail mailing list   Procmail homepage: http://www.procmail.org/
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail
<Prev in Thread] Current Thread [Next in Thread>