From: Dumas Patrice ([email protected])
Date: Mon Aug 28 2000 - 15:41:38 CEST
Sven Conrad a écrit :
>I find it confusing, if all Interfaces are defined in one file. Think
> about
> one point: eth get a IP out of one for the subnet, it is connected to,
> plip
> gets a point-to-point pair ppp most the time a automatic IP. It could be
> very confusing for newbys to get mixed questions.
IMO there are 3 things to do, here.
* The definition of the interfaces,
* the storage of the configuration files, and ,
* the bringing up and down of the interfaces.
from now on, there can be just one interface stored at one time, and there
is a script in setup (network.fun) that does the 1rst thing, the second, and
some things for the third (like module loading), and another script that
does the 3rd thing (netconfig), bringing up interfaces with ifconfig, and
starting or stopping routing.
> On the other side, I
> played
> around at some time with eth, plip and ppp (allso direct link) on a
> system.
> This means 3 network devices running at one time.
> With separate setups, there is no problem in doing so. How would you do
> this,
> if it is one file? One the other side, if this one script is avanced
> enough,
> may be the setup for multi eth is solved as sideeffect.
There is no reason why it couldn't be solved (even more easily) with one
.fun for each interface type.
>
There is another problem as I said in a previous email, that is that some
functionalities can be redundant. With plip, you can define a network, for
example, so there is some code that is redundant, and also in bringing up
and down interfaces. My idea is to define a script that should be sourced
(with . ) by the other .fun, so that they can share some functions
(something like network.utils which would in fact contain functions from
netconfig and network.fun).
But surely, a .fun for each interface type, with the possibility to define
several interfaces of the same type should be the better way.
> Allso, if there
> is
> more than one network device, the routing becomes more important. May
> be, this
> is a good centralized think. Any device could be set as direct or
> masquraded
> and stuff like this. That would be allso a solution for my general
> problem
> on plip, where I setup allso masqurading, which will interfere with ppp
> masqurading.
Maybe this is a different problem, that should be solved in the
corresponding .fun. The first thing to do in respect with this objective, I
think, is to define a way to store the different interfaces so that muLinux
could automatically detect the interfaces which are defined and the
masquerading could be centrally-defined.
Pat
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
This archive was generated by hypermail 2.1.6 : Sat Feb 08 2003 - 15:27:15 CET