On Mon, 19 Feb 2001, Phil Pennock wrote:
> We'd like to use an external command's exit status for triggering a
> director. Yes, we know about the potential overhead. :^(
Aarrgghh!!
In Exim 4 you might be able to do this using queryprogram (whose revised
spec I have not yet determined). Well, not this exactly, but maybe
enough to do what you want.
> Short of embedding Perl and using a perl function which does system(),
> is there a way of doing this?
No.
> Failing that: Philip, how much hassle would a
> ${command{pathname}{argv0}{argv1}...}
> expansion item be, please?
Lots, I suspect. Although I suppose it isn't much worse than ${perl,
come to think of it.
An easier spec would be
${command:text of command}
with the text being split up into args just like a pipe command. There
is no chance of this in Exim 3 now.
> On 2001-02-19 at 20:22 +0000, Dr Andrew C Aitchison gifted us with:
> > ${command{pathname}{{env0}{env1}...}{{argv0}{argv1}...}} ?
> > Or can anyone see a better way of controlling the environment ?
> >
> > I'm tempted to say "security", but I can't justify it.
>
> Exim already nukes most of the environment.
When it runs a sub-process, yes (see spec of "pipe" for example).
> Perhaps if the command is run under the uid of the invoker, if the
> invoker is not Exim's own UID or root?
You are talking about using this in a director. During delivery, the
invoker usually is exim or root (after message reception, or in a queue
run).
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
