[Ns-developers] [ns3] Statistical framework (draft)

[Tom Henderson]

> 
> Vincent, have you drafted code or more ideas?  I am not super-tied to
> the draft code in [1], but wanted to get something out there as well
> as demonstrate the kind of generic statistics I'm envisioning.  Your
> input on the approach as well as your help making this a generally
> usable component would be greatly appreciated.  Mathieu's writeup also
> provides some solid motivation, discussion, and references that should
> be followed up on.
> 

Joe (also Vincent and Mathieu),

Do you think you can converge on an overall API and use cases (and 
typical workflows) and document on the wiki?  I just read through this 
thread from Vincent's original post and it is not clear to me that 
everyone is in sync with respect to how they anticipate the system to 
work (or relative priorities).  And I'm sure there will be other 
priorities expressed.

Here are a few initial questions I had:

- are we going to build a system that is intended to hook only into our 
tracing framework, or one that can also be explicitly hooked into a 
simulation directly by editing the core?
- can we borrow APIs from other past statistics efforts?  Have people 
(other than Mathieu) reviewed these in detail?
- are we (and if so, when) going to work on a framework that exists 
outside of the main() program to coordinate multiple independent runs, 
or is that done from within main()?

Perhaps a specific question for Joe-- what would the main program look 
like for someone who wanted to implement a Counter that took as input an 
existing TracedInt (available via a trace path), counted the number of 
times that the value was positive (or some other arbitrary filtering of 
the output of the trace source), and stopped the simulation when the 
count of such events reached some specified number?  The reason I ask is 
that it is hard for me to anticipate what that would look like based on 
your current example that hooks directly into a repeatedly scheduled 
event-- it seems like some glue would be needed between the 
StatCalculator and the trace source.

Thanks,
Tom