[Olsr-users] Scalability of Networks using Olsr_Switch

[Bernd Petrovitsch]

Hi!

On Don, 2010-12-16 at 12:36 -0500, Robert Keyes wrote:
[...] 
> Reading this makes me wonder about the barely-utilized and non-tested 
> parts of olsrd. One maxim for quality software is that of "Ockham's 
> razor", meaning the simplest solution is most likely to be the true one. 
> Is too much 'cruft' accumulating in olsrd? Perhaps only so much as is 
> needed and tested should be included in the 'stable' branch.

Not that I ever had a different opinion, but the removal of allegedly
unused features, plugins, ... has the risk that people actually use it
but no one knows - especially if you are not there from the start.

a) I asked (years ago - late 0.4 and early 0.5 time IIRC) at least two
times who is actually using which plugin and didn't get a single reply
(granted, I knew of the one used in the FunkFeuer net so some of them
never were on the "very probably not used" list).
b) People in the mesh world do not update - for years! - (because people
are lazy in the whole world and mesh routers tend to be mounted on a
roof or attic or God knows where - and thus not that easily accessible).
Period. They update something if it's broken (and not even then always)
or they need/want desperately a new feature.

Take a) and b) together that I could remove an unused plugin today and
could get complains in 2-3 years from now (including the attitude that I
put it back in, forward port it and test it immediately).

Would you take that risk?

> What are the test procedures for each new release? Is there a network 
> testbed? Or a script which tests each option configuration to see that it 
> doesn't crash or go crazy?

The generic answer in the open-source world is: Please do and report any
findings. 

And don't forget the Cygwin, NetBSD, FreeBSD and OpenBSD port.

Bernd
-- 
Bernd Petrovitsch                  Email : (spam-protected)
                     LUGA : http://www.luga.at