On Sat, 30 Sep 2000, Adam Glass wrote:

> Unfortunately, about 50% of the time, the call gets disconnected *right*
> after the connection sequence completes, and right when getty is
> supposed to send the issue file.
> 
> On a hunch, I guessed that this was the result of getty sending the file
> too soon, interrupting the not-quite-completed connection.  Lacking
> sufficient experience to debug the source code vis-a-vis CD/DSR signals
> and such, I simply put a one second pause into the getty source at the
> point where it's going to send the "if" file.  (Is there a way to do
> this without modifying the source?  I couldn't find a way.)  Anyway,
> inserting a one second pause made the login sequence work correctly 100%
> of the time -- no more premature disconnections.
> 
> Has anyone else experienced anything like this?  I searched the archives
> but didn't find anything...

I could never get getty to answer the phone reliably; I just use
mgetty. Lot's of folks have reported following the same
course. That'll be cool if you can fix NetBSD's getty.


Frederick