On Wednesday 28 September 2005 22:50, Steve Lamb wrote:
> Adam Funk wrote:
> > That tenet is respected in the Unix tradition by using distinct tools
> > that have well-defined, debuggable interfaces between them (e.g. SMTP
> > and sendmail's handling of stdin).
>
> Ah yes, SMTP. And when that fails. Like the network isn't
> connectable? Do we just throw the message away. "Sorry, the remote
> queue isn't available, we'll just /dev/null it and have you guess at
> why your cron job didn't work. G'luck!"
No, we queue it locally in the MTA queue until we can shift it from there.
> > Without it, the batch job mail will be from and to
> > userid@???, which is probably not the
> > right address for receiving it or for replies and bounces.
>
> Call me crazy but I generally configure the programs so mail is
> deliverable to the proper person. 10 years, not one address rewrite.
> Address rewriting isn't need to do that, it is only the most difficult
> and needless way to do it.
If you prefer to do it that way, it's a matter of personal taste, but I
find that for cron and at, rewriting and using ~/.forward are the easiest
way to generalize the mail handling -- rather than configuring it over
again in each job.
