Re: Connection via modem

[Mark Eichin <eichin@cygnus.com>]

<lecture>

SLIP and PPP are closely related in some ways, vastly different in
others.

SLIP is simple to configure -- for a single leaf node, you need to know 
	your IP address
	their IP address
	the link MTU
	whether Van Jacoben header compression is on or off (CSLIP)
	their phone number.
That's it -- 4 numbers and a yes/no question. If you get that right,
you're on the air. Simple. Nothing happens "under the covers", that
really describes it all.

PPP, on the other hand, has automatic configuration, so that in
*theory* all you need to know is
	their phone number
everything else can be "negotiated". The flip side, of course, is that
if anything goes wrong, you've got lots of ppp logs to trace through,
and some subtle interactions... but if it works, it's easier.

PPP also a few options to handle things slip doesn't do at all:
	data compression (not the same as header compression)
	quoting (if you're running through something with flow
control, or otherwise not an 8-bit-clean link, ppp can handle
"protecting" those characters; SLIP can't.)
	support for other protocols (you can run DecNET IV over PPP,
not over SLIP. I'm sure you find that tremendously exciting :-) 

Because of the added complexity, there were times in the past when
linux ppp didn't work very well. There's also an issue of having the
right version daemon for a given kernel. On the other hand, things
seem ok now (and have for a year or more :-) The one CSLIP link I run
is between two radios, home to office; I control both ends, and
fallback to phone. Given the time to set it up again (which I may get,
the office is relocating soon :-) I'll probably use PPP...

</lecture>