<<< Date Index >>>     <<< Thread Index >>>

Re: mutt writes Fcc: even if the MTA fails to deliver



* Michelle Konzack <linux4michelle@xxxxxxxxxx> [2004-08-12 12:06 +0200]:
> Am 2004-08-12 08:52:35, schrieb Nicolas Rachinsky:
> 
> > I'm not sure wether I understand you correctly. What does your 'which'
> > refer to?
> > 
> > a)
> > No, it can't in David's above suggestion. If writing to the fcc happens
> > after sending the mail, the former can fail. And no, you can't detect
> > this before sending the mail, since the disk can fill up during (e.g.
> > because of) sending the mail.
> 
> But if the MTA fail to deliver and it does not queue, it 
> sends a message back to mutt that the delivery has failed. 

Obviously I wasn't as aclear as intended. The problem exists in the
case the sending of the mail is successful, but mutt cannot write to
the fcc.

-user presses <send>
-mutt starts sendmail, which successfully takes and delivers the message
 e.g. ssmtp hands it over to the smarthost
-<disk fills up>
-mutt cannot write to fcc

And now you have sent a message _without_ having a local copy.

And IMO this must not happen, if it can be avoided.

> some MBytes and some times my ISP's smtp-relays deny for 
> some time the sending of such big messages... So I do not 
> like to store a dozen of 10 MByte Messages on Disk or in 

Perhaps it should deny such mails all the time.

> > b)
> > No, if writing to the fcc fails, mutt should not (and does not) send
> > the mail.
> 
> If mutt fails to write Fcc then you have much mor problems...

Not in all cases.

> I think, a new mutt option which allow whether Fcc is 
> writen bevore send() or after it is send is neccesary to 
> fit the needs of all mutt $USER

Show us your patch. That would be something I could live with.

Nicolas