On Thu, 12 Jul 2001, Sheldon Hearn wrote:
>
>
> On Thu, 12 Jul 2001 14:40:12 +0200, robert rotman wrote:
>
> > I don't understand your question?
>
> You seem to understand my question well enough, you just don't seem to
> understand why I want what I want. That's fair enough, since I
Yes, english is not my first language:(
> neglected to mention in my original message that this is of academic
> interest to me. :-)
>
> > an 4xx error is a good choice for your problem.
>
> That's where you're wrong. :-)
>
> What Exim does right now is this:
>
> 1) If an lsearch fails because the key is not found, the expansion is
> null or <string2> after expansion. I've ommitted <string2>.
>
> 2) If an lsearch fails because of a "system failure" (e.g. open()),
> the expansion fails, breaking the caller (smtp_banner in this case).
>
in my opinion the question shold not be "how can i hide system failure
with my software", it should be "who can i avoid system failure".
exim can assume here that all your lookup-sources ar ok and
you have to care that they are stable at runntime.
>
> I want to know whether there's a way to control the way Exim handles a
> "system failure".
>
> I don't really care about the smtp_banner. Looking at the Exim spec,
> there doesn't seem to be a way to cause a system failure in a lookup to
> evaluate to null or <string2> if supplied. I just want to know whether
> I've missed some way. :-)
>
> Ciao,
> Sheldon.
>
robert
