[Ns-bugs] [Bug 631] RealtimeSimulatorImpl not compatible with python bindings

[code@nsnam.ece.gatech.edu]

http://www.nsnam.org/bugzilla/show_bug.cgi?id=631


Mathieu Lacage <mathieu.lacage at sophia.inria.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mathieu.lacage at sophia.inria.
                   |                            |fr




--- Comment #4 from Mathieu Lacage <mathieu.lacage at sophia.inria.fr>  2009-07-13 03:26:08 EDT ---
I think that a better solution here is simply to declare Simulator::RunOneEvent
+ Simulator::IsFinished dead and force GUIs to use a thread to run ns-3:
RunOneEvent is extremly complex to implement for anything but the default
simulator implementation (i.e., parallel simulators are stuck).

If we required the use of threads, we need to figure out how to allow users to
still interact with the simulation from another thread. i.e., we need to have
some amount of thread-safety built-in. One possible commitment we could make
would be to make the simulator.h header fully thread-safe (or, mostly
thread-safe) to allow a GUI thread to insert events in the simulator to perform
queries in the simulation objects from the simulation thread and then return
them to the GUI through a user-specific queue.

The only issue I feel concerned about is that of performance. What is the cost
of locking a simulator-wide mutex on event scheduling micro-benchmarks ?


-- 
Configure bugmail: http://www.nsnam.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.