Re: [ns] satellite+wired simulation

[Lloyd Wood <l.wood@eim.surrey.ac.uk>]

On Thu, 21 Dec 2000, Tarik Alj wrote:

> I might not have anything to do with satellite or wireless stuff
> in NS, but I do agree with that sentence. My point of vue is that
> if we are going to play with low level stuff, I mean below L3
> (e.g. LL, MAC, Phy, Channel), we ought to provide for "real"
> interfaces and links. LL + MAC should just be seen as an interface
> and phy + channel as link; and nodes should suport heterogenous
> interfaces. The current model allows nodes to be either wired or
> wireless, does not allow a node to have different interfaces and
> is just plain UGLY (yuk-yuk)  in my opinion.
> 
> I would like to know your opinion on that. Lloyd?

Oh, absolutely; there seems to be a lot of object flexibility not
being taken advantage of, mainly because (as far as I can see)
different bits of code are frozen at different types of class API.
Configuring nodes when you create them with a list of limited
options really isn't really that flexible, as opposed to installing
interfaces etc.

Tarik, I look forward to seeing your rewrite of the code. (can you
make all other addressing types derived forms of hierarchical
addressing and rewrite the multicast code to work with wireless
routing while you're at it?)

seriously though, it _is_ a mess; rewriting would require throwing
out/porting a lot of functionality (throwing away stuff might be the
only way you'd get a better simulator - ns v3?), and is, alas, a
mammoth thankless task. Who'd do it?

L.

<L.Wood@surrey.ac.uk>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>