>> ... I teach them about the alias so they know what
  >> is going on. If they man rmi or cpi they will not
  >> find it so that is a much worse idea then just a
  >> bit of explanation.

  der> "rm *" in a directory and expecting to be able
  der> to say yes or no to every file will not work.
  der> "ksh: rmi: command not found" is easier to deal
  der> with than lost data.

I'm with Hans on this one. In 19 years of using Unix, I
could probably count on one hand the number of times
I've used the -i flag to rm. Or maybe on two hands.

The correct behavior of rm is not to ask the user for
confirmation. That's what they should learn, and also
how to deal with it. 

It infuriates me when someone changes the defaults for
me. I remember when I first started using Red Hat it
frustrated me that no matter what I did it would ask me
for confirmation. It was so far beyond my ability to
comprehend that someone would have the audacity to
change the behavior of a Unix command that had worked
fine the way it is since 1969 that it didn't occur to
me until sometime later that RH had aliased it. A big
fat boo to them for doing that.

Users who want that sort of behavior are entitled to
create it themselves.

Somehow it reminds me of when I drove a U-Haul from the
coast of Maine to Arizona in 1978. The truck had a
governor on it that prevented me from going faster than
40 miles an hour with the pedal to the metal, maybe 45
if I was going downhill. Presumably someone at U-Haul
thought they were doing customers a safety favor by
forcing them to drive 15 miles under the speed limit
all the way across the country.

-- 
Lynn David Newton
Phoenix, AZ