Feedback on MailScanner 4.56.1-1 beta

[mikea]

On Fri, Aug 25, 2006 at 11:45:47PM +0200, Jim Holland wrote:
> On Fri, 25 Aug 2006, Glenn Steen wrote:
> 
> > On 25/08/06, Julian Field <mailscanner at ecs.soton.ac.uk> wrote:
> > >
> > > Denis Beauchemin wrote:
> > > > Jim Holland a écrit :
> > > >> On Fri, 25 Aug 2006, Julian Field wrote:
> > > >>
> > > >>
> > > >>> That is what I would expect to happen. RPM won't overwrite a file
> > > >>> owned by another package unless forced to (which I don't like doing).
> > > >>>
> > > >>
> > > >> I understand that.  My original question was why once the install.sh
> > > >> script experienced a problem with one module it would then fail to
> > > >> attempt
> > > >> installing the remaining modules.
> > > >>
> > > >>
> > > >
> > > > Not on my RHEL 4 system.  It tried to install them all.
> > >
> > > Thankyou for that. I'm not going nuts after all. On everyone else's
> > > systems it tries to install them all, it doesn't give up after one
> > > failure. It carries on and tries to install the next one.
> > >
> > I wonder if this has something to do with the fact that Jim is doing
> > his install on RH 7.1 ... Somewhat out of date, one might say (even
> > with legacy updates:-)
> 
> It is also driving me nuts too - not so much by the problem, which I can
> cope with, but trying to understand what is going on!  I am totally
> baffled.  Even if it is an ancient OS, a bash script shouldn't bomb out if
> a command returns a non-zero return code or even fails altogether.  I have
> cut out the loop itself and run it without the rpm command, and then it
> works fine, going through each package and making appropriate comments.

This came up on another list I'm subscribed to, and yes, indeed, some 
shells _do_ abort a script on a non-zero RC or if a command in the 
script failed. This caused a lot of heartache and head-examination. 

There is no good, universal solution, but it probably would be fairly 
easy to cobble up a short, guaranteed-to-fail script for folks to try 
as part of their next MS installation. It would be nice if it kicked 
out a warning on failure, or if it were written to be silent when it 
failed and to write a big, gaudy rooster-crowing success message. 

> I will let people know if I ever manage to get to the bottom of it.  For 
> now it is a bit like a crossword puzzle that you can't finish but keep 
> going back to.

What about changing the #! line at the head of each script to a shell 
that you know tests good on your system _and_ supports Jules' scripting 
needs? 

-- 
Mike Andrews, W5EGO
mikea at mikea.ath.cx
Tired old sysadmin