[Sussex] Project Advice....

[Richie Jarvis]

Hi All,

I was thinking on the way home (yeah, I know fatal.....)

As some of you know, I help run a wifi mesh club around here, and we 
have a problem (well, its a perceived problem, but a problem 
nevertheless!)  I am hoping to get some advice on how best to approach 
this problem within the ideals of the open-source community from you all....

Right, so the problem is this - when we started about 2 years ago, the 
only available wireless mesh software available for free was 
Locustworld's MeshAP.  Its Linux-based, but is not open-source.  It is 
also tightly integrated to WIANA (Wireless Internet Assigned Numbers 
Authority - wiana.org.)  Basically, for those that aren't familiar with 
this setup, WIANA provides 1.x.x.x addresses to each wireless node in 
the mesh network via a registration system, and also provides a 
remote-management system for the nodes, so that you can remote-load 
settings to them, and see whether they are up, down, or sideways.

It all works pretty well most of the time, but has a few niggles which 
really annoy me (and several of the others in the group, and elsewhere 
in the world that are using this system.)  So what I want to do, and 
have done for ages, is to write our own version of this system, which is 
compatible with the AODV Mesh (for easy switchover), can be loaded to 
the nodes without having to remove them (the MeshAP can download and 
install new versions in-situ, by overwriting the cramfs files), and can 
interface with the WIANA system (pretty easy, WIANA downloads a 
text-based key:variable file, and then a script runs through it all to 
setup the node itself.)  Oh, and the other thing that would be great is 
if we can also produce a similar setup on the OpenWRT system, that can 
inter-mesh with the old compaqs we have, as this would reduce our node 
costs considerably.

What I need is some advice here.  I am worried about doing all of this 
myself - I've been very tempted to try and do it all myself many times, 
and have not got past the thinking and requirements stage as yet.  I am 
also scared that if I do try and do it myself, I will make a hash of it, 
and leave some crucial flaw in the system.  The upshot of all of this is 
that I need to better understand how to sell this to the open-source 
community, and where it really fits in the grand scheme of things, as 
this would be something which (a) I am interested in, and (b) I need 
some help to implement.

I think that there are enough mesh networks around now who are using the 
MeshAP, but would jump at the chance of something which is truely open.

Another of the drivers here is that we (SussexNetShare) recently got a 3 
month trial of LocustWorlds paid product through my fellow co-founder 
(he hounded them), the MeshAP Pro, which is supposed to solve alot of 
the problems with the MeshAP, but in reality, it seems to be more of the 
same :(

This email turned out alot longer than I expected - sorry about that.  
If anyone has any pointers as to where I go from here, to a newbie in 
starting an Open source project, that would be great!

Cheers,

Richie