On Fri, 9 Dec 1994, Patrick wrote:
 
> Opinions will differ, I'm sure, but from here it looks like "auto-delete"
> is one of those  "improvements" that must have looked  good on  paper but
> turns out to be a pain in the neck in practice.
 
Well, I haven't  used it, but I  can see that it  could cause  problems.  I
still think auto-delete's still  a good idea  though -- maybe it just needs
to be tweaked a little.  People who use SmartList, another list manager for
Unix, rave about its autodelete function:  like LISTSERV, it first tries to
determine  whether  errors  are  "temporary"   or  "permanent," but  unlike
LISTSERV,  when  it  gets a  "permanent"  error,  rather  than  immediately
deleting the address  it simply starts  a counter of consecutive  errors on
mail to  that  address.   When the  count   hits some  list-owner specified
number, the server sends  a final warning to the  account (which may bounce
of course, but at  least it tried).   Next  bounce, it deletes  the account
while notifying the list owner.  How does that sound?
 
I think auto-delete is an important productivity feature for big lists, and
it's great that LISTSERV has it.  Still, everything can be improved.  Eric,
if you're  there,  what  do *you* think?     Am  I missing  some  important
implementation  issue?  Could some  sort  of "tolerance"  be built into the
auto-delete function?  I think if I  were designing it I'd prefer something
like "x unsuccessful deliveries, but don't delete until  at least x days of
no success," just in case some heavy  wave of mail  pushed someone over the
bounce limit in a very short time ...
 
Norm