Path: utzoo!attcan!uunet!lll-winken!xanth!ukma!tut.cis.ohio-state.edu!bloom-beacon!EXPO.LCS.MIT.EDU!keith
From: keith@EXPO.LCS.MIT.EDU (Keith Packard)
Newsgroups: comp.windows.x
Subject: Re: xdm question
Message-ID: <8904061523.AA21681@expo.lcs.mit.edu>
Date: 6 Apr 89 15:23:44 GMT
References: <34700009@primerd>
Sender: daemon@bloom-beacon.MIT.EDU
Organization: The Internet
Lines: 29


> I am trying to use xdm to control a couple of X terminals.  
>
> .... but the problem seems to
> be that once xdm has made the connection, it doesn't realize when the server
> goes down.

This is because xdm doesn't send any packets to the display after the
connection is established, so the network connection remains idle and xdm
never finds out that the other end has disappeared.  You could use
the TCP KEEPALIVE stuff from BSD and hope that works correctly, or
hack xdm to call Xsync every 5 minutes or so.

As with almost all other terminal issues, this one is easily resolved
with XDMCP.  Until such a protocol exists you'll probably want to leave
the terminals on-line all the time.

When you do want to turn the terminal off, you should do so before xdm
gets around to opening the display for the first time.  The open will fail
and xdm will disable the display.  When you turn the terminal back on,
send a SIGHUP to xdm and it will re-read the Xservers file, notice that
the disabled display is still in the file and attempt to reenable it.

A cheap hack, but effective (i.e. I use it occasionally).

						Keith Packard
						MIT X Consortium
						(617) 253-1428
						keith@expo.lcs.mit.edu