🚀 go-pugleaf

In article <C1H53I.LuF@oceania.com> kambiz@oceania.com (Kambiz Aghaiepour)  
writes:
> 
> I am having trouble -NXHost'ing an application from a client machine  
that is
> connected to my network via slip.  (Yeah, I've heard it's really slow  
and
> not worth the trouble; but I enjoy pain!)
> 

You don't enjoy that much pain, but... If you examine the /etc/rc* scripts  
and perhaps the boot messages, you'll find that the Mach net message  
server looks around when it starts to see what network interfaces are  
available.  Since this is before Slip is loaded, it doesn't know about  
slip.  Thus you must somehow establish routes so that when the  
netmessageserver tries to send through the ethernet interface, the packets  
actually get routed correctly.  There are various approaches; for example,  
give the ethernet and slip interfaces the same IP address, then on the  
standalone machine in /etc/rc.local:
/etc/route delete 128.2.0.0 $INETADDR
Replace 128.2.0.0 with your network.  

Getting NXHost working over slip is an interesting experiment, and you can  
even get windows up on your screen if you go about it right, but it  
doesn't *really* work.  I think that what happens is that the interaction  
between the remote app and the windowserver consumes essentially all of  
the communication channel just to blink the cursor in a text object, and  
once you try to do anything more, buffers fill up and messages don't get  
through.  (This is at 19.2kbps, headers not compressed.)

Maybe if you just need to push a couple of buttons in an app, you could do  
something useful with NXHost over slip, but the moment you must type  
anything, forget it.  It Just Doesn't Work.

And by the way, you don't have to log out and back in after checking  
Public Window Server in 2.1.  Is that yet another "improvement" of 3.0?

 -- Derek
-- 
Derek_Beatty@cmu.edu   ABD   Comp Sci, CMU, 5000 Forbes, Pgh, PA 15213 USA