[openbeosnetteam] Re: PPP & so on...

> > > OK, having read through some of the RFC's, here's what I propose 
> > > we
> > > do...
> > >
> > > I am concerned that we'll be unable to test PPP much without 
> > > suitably
> > > easy
> > > test environments beyond a few people who happen to have PPP 
> > > setups.
> >
> > Sounds like a decent idea. Of course, a substantial number of 
> > people
> > have PPP setups, like me, for instance.
> > > So, we start by adding a serial line module. We should then be 
> > > able
> > > to write
> > > a ppp module that can do ppp over serial (RS-232) connections. 
> > > This
> > > should
> > > make testing easier.
> >
> > Actually, it's a lot easier to start with PPPoE. It requires no
> > complicated encapsulation, I use, and have code for it. I also have
> > HDLC code, but (1) it's less robust, and (2) HDLC is a hell of a 
> > lot
> > more complicated, and I think we should start with the simplest 
> > case
> > first.
> 
> True, and PPPoE support will sort of start appearing...

Sounds good :)

> > > Then we do the ppp module, which will require an LCP, IPCP and 
> > > CHAP,
> > > PAP
> > > modules to be written.
> >
> > Yes.
> >
> > > I also propose that we add a pppoe module as this should be
> > > relatively
> > > straightforward to do. I see a problem in trying to do too much 
> > > at
> > > once as
> > > bugs will creep in and be worked around in one module or another, 
> > > so
> > > we take
> > > it a step at a time.
> > >
> > > 1. Anyone volunteer to do a simple module that can open a serial
> > > device and
> > > read/write to it?
> > > 2. we do basic PPP encapsulation code.
> > > 3. we do LCP module
> > > 4. we add state control to ppp module
> >
> > State control is part of LCP.
> 
> Hmm, well yes and no. PPP needs to know about the state whereas it 
> gets
> changed by LCP...

Of course, you could let LCP know about the state, and then call things 
itself, but that distinction makes sense.
 
> > > 5. we add chap module
> > > 6. we add authentication handling to ppp module
> >
> > CHAP *is* a form of authentication. Why are we adding auth *after*
> > CHAP? Also, where is PAP in all of this? Very, very few services 
> > use
> > CHAP, you know, so I'm not sure this should be an immediate 
> > priority.
> > (It's easy enough to do, at least with MD5 CHAP, but PAP is still
> > infinitely more common)
> 
> Yes, I mean PAP and CHAP. Why CHAP? A lot of people want it (or at 
> least
> most people who email want it).
> 
> The reason is we add the module, then we go back and add hooks to use 
> it. I
> think it'll be possible to test the PPP stuff outside the server and 
> that's
> what I intend to do. I don't want to hook into the auth modules until 
> we're
> sure they work.

Ok, cool. I had thought you didn't intend you do PAP, which would have 
been colossally stupid :P. I think I will hack the ethernet module to 
start playing with PPPoE, and begin building a PPP stack on top of that 
this weekend.
-Nathan

--
Fortune Cookie Says:

How come wrong numbers are never busy?

Other related posts: