On Wed, Aug 24, 2011 at 08:03:41PM -0400, Santosh S Vempala wrote:
Sven: Thanks for scrolling through the code. Have you actually tried it? Do you have any improvements to suggest? (even if you're not an expert at kernel programming)
I will _not_ try to use the code before the obvious problems aren't solved (make it thread safe, don't access memory and then check if it was good to access them, don't implement your own linked list, general error checking, don't clutter up proc, fix skb handling/checking in general, ...).
But don't understand me wrong. I think it is not a good implementation, but the idea itself is interesting and I would like a real paper/journal article about it. I know that there are at least two other papers available [1,2], but only one [2] is linked at your site. So maybe they don't show your current ideas... but I will read both of them later in the train.
I personally don't understand Jon Roland's comments. It is good that research is done in this field and also that people think about different goals they want to achieve. And I would also aggree with Marek that it would be interesting to show your results at small mesh related meetings like Wireless Battle of the Mesh (not real to "battle against other meshes", but to exchange some information). But be aware that the worst participant is eaten by wild animals ;) (just look through the pictures of WBMv4[3]).
Kind regards, Sven
[1] http://conferences.sigcomm.org/hotnets/2007/papers/hotnets6-final57.pdf [2] http://dritte.org/nsdr09/files/nsdr09_camera/s4p2_paranjpe09nsdr.pdf [3] http://battlemesh.org/BattleMeshV4/Pictures