I'm incredibly tired right now, so please forgive the incoherent thoughts
and idle speculation.
On Thu, 12 Nov 1998, Cary B. O'Brien wrote:
> What I had in mind was one slip connection to which gets routed all
> the ip addresses for the remote sites.
> When a packet comes in, diald looks at the target IP, then figures out
> which chat script to run and then which ppp options to pass pppd.
This is almost possible now. Diald passes information (MODEM and FIFO) to
the chat script by environment variables. It could conceivably pass the
IP number that started it up, too. (Thanks for the suggestion, Jim! :-)
I don't know what pppd would require from diald. I assume nothing other
than what it gets. As the link layer, all it really needs to know are the
identities of the two machines on either end of the link. Keeping it
simple would be nice.
> I guess then pppd would bring up the ppp interface and add the route to
> the site. Overall routing would see the ppp interface as being a better
> match and use it instead. Someone would have to look at timers.
No, diald messes with the routes. I think the addroute directive can
already be used to invoke a script with enough information to add the
desired route, and leave the default route pointing to the slip device.
> There would only be 16 or so modems, so there would never be more
> than 16 pppds running.
> At this point this is something rather different from diald, but might
> use some of the same code, i.e. setting up the fake slip port, and
> looking throught packets for headers like the rules stuff does.
Hmm, not so much different. I think the only requirement other than
passing the IP destination to the chat script by environment variable is
that diald fork when it starts the connection process, so that one diald
can continue to listen on the default device while the other listens on
the new device (so it knows when to shut down the link).
> One would have to worry about race conditions if incoming and outgoing
> calls happen near the same time. I guess the diald mechanisim for
> incoming ppp calls could be used here to.
Incoming calls? This sounds like a pretty hefty application you have in
mind.
> NB: I have not looked at the diald code in a while. This is purely
> conjecture at this point. Still, I may need it in a while.
Me too! :-)
Ed
--
Ed Doolittle <mailto:[EMAIL PROTECTED]>
"Everything we do, we do for a reason." -- Peter O'Chiese
-
To unsubscribe from this list: send the line "unsubscribe linux-diald" in
the body of a message to [EMAIL PROTECTED]
