[ptp-dev] Debug / Runtime interface

[Nathan DeBardeleben <ndebard@xxxxxxxx>]

Donny, Clement - how are you guys envisioning the interaction between 
the debug and parallel models?  I'm not entirely clear what happens when 
someone clicks on the 'debug run' button as opposed to the 'run' 
button.  I believe a configuration is passed into the same run method 
but that configuration has a debug flag set - is this correct?

We can easily change the interface to the runtime system (control system 
more specifically) so that the Run() method takes as argument a flag 
which is true if debugging is enabled.  But - how do you intend to start 
the debuggers?  One thought I had for modeling this might be:

ModelManager.run(...., required_debug[true/false], ...)
{
   if(required_debug == true) {
      debugModel.run(controlSystemHandle, ...);
   }
   else {
      controlSystemHandle.run(...);
   }
}

Then if the debugging was required, you could do whatever preparation 
you needed for the control system in the debug model.  I could hand you 
a handle to the control system so that you could call 
controlSystemHandle.run(...) whenever you were ready (after preparing 
the debug system).

These are just some ideas - I'm interested in hearing what you guys are 
thinking about.

--
-- Nathan
Correspondence
---------------------------------------------------------------------
Nathan DeBardeleben, Ph.D.
Los Alamos National Laboratory
Parallel Tools Team
High Performance Computing Environments
phone: 505-667-3428
email: ndebard@xxxxxxxx
---------------------------------------------------------------------