>On May 2, 2008, at 4:33 PM, Paul Stickney wrote:
>> Hello,
>>
>> TOSSIM-T2 seems to have undergone a complete transformation from the 
>> TOSSIM present in the TOS 1.x line.
>> In particular, TOSSIM-T2 (and Python) now appears to work directly 
>> through a C++/SWIG interface while it (TinyViz, Tython) had a JVM 
>> interface.  While TOS 2.x changes appear to be a *good* thing, TinyViz 
>> (and the related plug ins) no longer seem to be usable in any extent.
>> I have been searching the mailing lists, but have not found sufficient 
>> answers and new solutions.
>>
>> 1a) Is there a nice (read GUI) program to interface with TOSSIM-T2 and 
>> interact with the network?
>No. The intention was that by having a Python/C++ interface, it would be
easy for someone to write one.
>
>When I wrote TOSSIM for T2, I explicitly avoided doing this, for 2 reasons.
First of all, any GUI-based >interface is inherently unrepeatable. Better to
have scripting (this is what motivated Tython in 1.x). Second, >getting
TinyViz to be useable required punching a few holes in the simulator/mote
boundary (such as >FakeLocation), which then led some systems and protocols
to depend on information which would never be possible >to obtain in a real
network.

I'm currently working on making a very simple GUI, written in python
(pyGTK), that provides the ability to run simple simulations.  It hopefully
will help people get started with TOSSIM but WILL NOT BE DEVELOPED FOR
COMPLEX SIMULATIONS.  To do complex simulations you will have to go back to
writing the simulation in python.  The GUI inherits the TOSSIM objects and
extends them with some basic features like x,y coordinates for the motes.
When I get a chance I'll write a quick abstract for the GUI and post some
current design notes.  As soon as I get a stable version up I'll post it.
Currently I have an unstable version working with topology for the motes,
radio communication, variable speed for running the simulation, log
filtering and the ability to view variables on each mote working.
http://www.cs.chalmers.se/~dcs/seminar/ is a link to what looks to be
someone else developing a GUI for TOSSIM.  They do not say what version
there developing it for.


>> 2a) Is there an easy way of providing feedback to motes while 
>> bypassing the radio stack?
>> In particular, I wish to "simulate sensor data". Looking through the
>> C++ driver, there is only an exported getVariable (I believe this
>> could be hacked as it seems internally pointer-based, but I'd rather 
>> not make additions that require patches.)
>Currently, there is not. I thought about incorporating a Python/C++
callback mechanism for ADCs, but the >overhead is pretty high and I wasn't
sure this was the right answer. If you're willing to go with C+ 
>+, writing your own accessor function is really easy: just write a
>TinyOS sensor module that calls a C accessor function, and write a mutator
function that your C++ driver can >call. Going through Python would require
using SWIG to export the call.

This is another extension I'm working on.  It is a new object for TOSSIM-T2
that will provide the ability to set a sensor value in your python script
and read it in your motes.  It will also read a stream from a file pointer
passed to it.  It is going to be very simple with get/set functionality with
NesC objects calling gets to read the values and Python sets to configure
the values.  I made it very specific so far making it only work with the
HamamatsuS1087Par sensor rite now.  I'm currently reading trying to find a
way to imbed it more with the ADC to allow for a more dynamic design for
more platforms.

Raymond

<<attachment: winmail.dat>>

_______________________________________________
Tinyos-help mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help

Reply via email to