[Ns-developers] ns-3 roadmap revision

[Tom Henderson]

Tom Henderson wrote:
> We are soliciting feedback on the design of the ns-3 node, and in
> particular on how node objects are bound together. I've posted a
> document at the below URL that summarizes the current open issue on
> node architecture. We'd like to discuss this issue on the list and
> take a decision by the Nov. 24 timeframe.
> 
> -Tom
> 
> <http://www.nsnam.org/temp/node.pdf>   (temporary URL for this month)
> 

A little while ago, when we announced this project's infrastructure 
(September), we posted an initial roadmap for development:
http://www.nsnam.org/roadmap.html

Regrettably, we are not keeping pace with that roadmap because we have 
had some difficulty coming to agreement on some issues.  Presently, I 
would say that we have met the September milestones, and we are working 
on the October milestones still, but it is now the end of November.

I would like to revise the roadmap.html and shoot for a new goal that 
would have us make a pre-alpha (i.e., not feature complete) release by 
early February 2007.  The new roadmap covers most of the milestones in 
the current roadmap, but maybe some more limited functionality for some 
of them.  Instead, the hopefully modest goal is to replicate ns-2's 
simple.tcl script in functionality:
- four node fixed topology
- CBR sender and sink
- FTP sender and sink
- UDP client and server
- TCP client and server
- static IPv4 routing
- serial (wired) point to point links between nodes
- FIFO, droptail queueing
- libpcap trace of all nodes
- limited command line argument processing
- limited statistics capability
- documentation of the above

Maybe we will get some other things done by then, but the above would be 
a tangible target.  I think that this relieves us trying to have a 
perfect architecture at first cut, instead focusing on getting some 
functionality.  I understood Lee Begg's feedback to be along these lines 
(focus on the important interfaces, and then see what architecture 
shakes out).  I also have heard sentiment that people are eager in 
general for some running code.

I personally am prioritizing the Mac/Network interface at the moment, so 
as to allow George to progress on upper layers of the stack, and Mathieu 
on lower layers in tandem.  If people generally agree on the above 
target, I'll suggest some additional interim milestones to get there.

- Tom

p.s. I'd like to thank Lee, Gustavo, and Lloyd for reviewing at least 
some of the sample code and providing comments on the list.