Subject: Re: clear screen when logout
To: Ben Collver <collver@softhome.net>
From: Richard Rauch <rauch@eecs.ukans.edu>
List: port-i386
Date: 10/20/2000 05:15:41
> Let me clarify: 
> I was only describing how "it was done" in Debian 2.2.  I made no value
> judgment, the (in)correctness is obvious.

I'm sorry for misunderstanding you.  I took the word ``solves'' to have a
narrow meaning (i.e., ``completely solves in a manner that would work
here'').  I wasn't really addressing (in)correctness, but rather
completeness---the Debian approach doesn't even count as an incorrect
solution (much less correct solution) to the desired feature, as I see it.

(``Solution to the desired feature?''  Yeah, that's a yucky way of saying
it, but it's not really a _problem_, per se, and I feel mired in the
``solution'' terminology here.  *grin*)


From the sound of it, I think someone has proposed what may be the best
solution: Use an existing feature of getty to run an arbitrary program
between logins.  But, I being only an egg, I'll comment-on/ask-about a
couple of your points below:


> One solution would be to extend getty so that termcap capabilities could be

Correct me if I'm wrong, but this will tend to be hosed if /usr isn't
mounted for some reason.  Either we dynamically link it (and presumably
fail to run getty if /usr isn't mounted) or we statically link it (and
suffer unsightly bloat).  (Unless, as you note, it already links against
libtermcap for some reason...)

While relying on the shell to do the right thing seems inherently a bad
idea, it also seems wrong to put this feature directly into getty.

Yes?  No?


> specified in /etc/gettytab ie. cl=tc#cl, but then getty needs to be linked
> against libtermcap if it isn't already, and the terminal type needs to be

(Maybe we're saying much the same thing, re. libraries, here...(^&)


  "I probably don't know what I'm talking about." --rauch@eecs.ukans.edu