[Ns-developers] tracing requirements

[mathieu lacage]

and, as usual, I forgot the actual attachement. So, instead, I pushed
the full doxygen documentation to my webspace:

http://www-sop.inria.fr/dream/personnel/Mathieu.Lacage/ns-3-tracing-doc/group__tracing.html

enjoy,
Mathieu

On Mon, 2007-03-12 at 16:19 +0100, mathieu lacage wrote:
> hi tom,
> 
> I have taken a stab at implementing some low-level code to make all of
> these requirements possible. The current code allows you to define
> multiple trace sources, connect an arbitrary number of trace sinks to
> each trace source. The current code does not attempt to provide the
> final solution where all you have to do is just specify a trace file.
> Instead, it deals with the problem of getting the trace events, and
> collecting them all in one place with enough information to generate the
> proper trace file. Sample code demonstrates how the actual trace file
> generation could be done based on the tracing information.
> 
> I have pushed all this code in
> http://code.nsnam.org/mathieu/ns-3-tracing
> 
> I have modified the two default sample programs
> samples/main-serial-net-device-if.cc and samples/ns-2/simple.tcl.cc to
> use this framework to generate some simplistic trace files. If you want
> to see what needs to be done in a model, you can look in
> src/node/queue.cc Queue::CreateTraceResolver.
> 
> Furthermore, I have pushed extensive doxygen documentation on both the
> high-level design and the detailed API. I have attached to this email
> the main documentation file but anyone should be able to re-generate the
> doxygen doc with the "scons doc" command and should be able to browse
> the tracing section (see doc/html/group__tracing.html after doxygen
> generation)
> 
> The current code allows you to trace all queue events in SerialNetDevice
> objects, all ipv4 tx, rx, and drop events. Many more could be added but
> I did not bother adding tracing support everywhere for now. It did not
> seem to bring much more features. The goal is mostly to demonstrate the
> idea that this proposal should be able to do all the things outlined in
> this requirements document.
> 
> As usual, comments would be welcome.
> 
> regards,
> Mathieu
> 
> On Fri, 2007-03-09 at 10:48 -0800, Tom Henderson wrote:
> > Here is a current snapshot of our assumed requirements for ns-3 tracing:
> > http://www.nsnam.org/temp/trace-requirements.pdf
> > 
> > We plan to implement a subset of these requirements for our pre-alpha 
> > release; namely, an ability to generate "out.tr"-like traces of the 
> > queues, and perhaps also pcap-formatted traces.
> > 
> > Comments, please,
> > 
> > Tom
> >