Suspend

[carothersc-zz]

Use this functionality by calling tw_lp_suspend in your event handler. An example is:

if( lp->gid == 0)
{
    printf("LP 0 is suspended at TS = %lf - should only get this once!!\n", tw_now(lp));
    tw_lp_suspend( lp, 0, -31459 );
    return;
}

[codecov-io]

Current coverage is 74.63%

Merging #75 into master will increase coverage by +1.13% as of bca71bb

@@            master     #75   diff @@
======================================
  Files           25      25       
  Stmts         2457    2456     -1
  Branches         0       0       
  Methods                          
======================================
+ Hit           1806    1833    +27
  Partial          0       0       
+ Missed         651     623    -28

Review entire Coverage Diff as of bca71bb

Powered by Codecov. Updated on successful CI builds.

[JohnPJenkins]

Are the event pointer / timestamp checks sufficient to avoid issues with event ties in optimistic mode? Not sure what the semantics regarding ordering of ties are.

[carothersc-zz]

Hi John - yes, absolutely - no other event will have both the same pointer and time stamp.

Note, the time stamp is used solely to check that the event has not been re-used since
the LP was suspended. That is you suspend an LP and it presists til the end of the simulation.
That event will be re-used many times -- in fact once GVT sweeps by, you know that LP
will be suspended til the end ..

Make sense ?

thanks for the code review ...
Chris

[JohnPJenkins]

That makes sense. Happy to help review, I was after all pestering for the feature :).

[JohnPJenkins]

The suspend flag needs to be unset here - otherwise, a suspended lp always stays suspended (and future RCs are never processed).

[carothersc-zz]

Thanks for the catch John.

There was a case where you don't want to unset the flag.

Let me read the code and double check that this is not that
case..

thanks,
Chris

[carothersc-zz]

Hi John - I just pushed the fix.
thanks,
Chris

[JohnPJenkins]

So I pulled in your suspend flag fix and re-tested the codes suite. The new suspend code works as expected!

[carothersc-zz]

Great news!

Thanks,
Chris

---

Christopher D. Carothers

Director, Center for Computational Innovations
Professor, Department of Computer Science
Rensselaer Polytechnic Institute
110 8th Street
Troy, New York 12180-3590

e-mail: chris.carothers@gmail.com
web page: www.cs.rpi.edu/~chrischttp://www.cs.rpi.edu/%7Echrisc
phone: (518) 276-2930tel:%28518%29%20276-2930

fax: (518) 276-4033tel:%28518%29%20276-4033

---

From: John Jenkins [notifications@github.com]
Sent: Monday, November 02, 2015 10:50 AM
To: carothersc/ROSS
Cc: Carothers, Christopher D.
Subject: Re: [ROSS] Suspend (#75)

So I pulled in your suspend flag fix and re-tested the codes suite. The new suspend code works as expected!

—
Reply to this email directly or view it on GitHubhttps://github.com/carothersc/ROSS/pull/75#issuecomment-153059385.

[gonsie]

closed #15