How nice should I be on exit?

[Vance Shipley]

Following the Erlang philosophy we let crashes occur when there
are things going wrong.  A process should die when it is abused,
this is normal.  It takes some getting used to though as we tend
to want to program defensively.

Now I'm asking myself whether I should worry about cleaning up 
when a process is going to die normally.  In this case I have 
set up monitoring of a number of other processes and stored the 
references.  It's a small pain to go through the list, pull out
the references and run erlang:demonitor/1 on them.  I'm thinking
that in all likelyhood it would be more effecient (oops!) to let
the run time system clean up after me.  The same could apply to
ets tables, ???, ...

What say the experts?

	-Vance