[certi-dev] Time management problem

[Евгений Чемерицкий]

Dear CERTI developers,

I have made a patch for testing some functionality related to HLA 
1516-2000 standard.
Currently it contains a problem associated with time management I cannot 
solve.
The simulation runs extremely slow and it seems to be a subject of time 
creep.
The number of NULL message exchanged at the end of either federate is 
too big (> 100 000).

The test includes two federates: bcvm (stands for central computer)
and one terminal (there could be more terminals but this test use only 
one).
Terminal has a counter with a specified initial value and sends 
interactions to bcvm,
bcvm replies, terminal decrements corresponding counter and
this cycle repeats until the counter reaches zero value.

Both terminal and bcvm are timeConstrained and timeRegulating 
simultaneously.
Model time is advanced by means of nextMessageRequest call and
federates are hanging for a long time during this calls.
Lookahead parameter value is rather big (lookahead = 10).

If you are interested by this problem or already have an idea of 
possible solution
please visit the patch tracker for more details
https://savannah.nongnu.org/patch/index.php?7487

Best regards,
Eugene Chemeritskiy