13:48 <hkaiser> gnikunj[m]: https://cdash.cscs.ch/viewTest.php?onlyfailed&buildid=140089

13:49 <gnikunj[m]> Strange. Let me look into it. It worked fine for me on rostam.

13:50 <hkaiser> this is on rostam

13:50 <gnikunj[m]> :/

13:50 <hkaiser> same on daint, btw

13:51 <gnikunj[m]> why is plain_async_distributed failing. It doesn't even use distributed resiliency APIs

13:52 <hkaiser> shrug, I have not looked

13:52 <gnikunj[m]> aah it's the assertion that's failing f_nodes < locales.size()

13:52 <gnikunj[m]> how many nodes is it running on?

13:52 <hkaiser> you may ignore those not related

13:52 <hkaiser> I set them to run on 2 localities

13:53 <gnikunj[m]> it should work then. f_nodes is set to 1 and if its run on 2 localities locales.size should return 2.

13:53 <hkaiser> I think I did :/

13:54 <hkaiser> ahh, got it, I didn't set the localities for the perf tests

13:54 <gnikunj[m]> 1d stencil code seems to be seg faulting. An indication that I updated the wrong code. I should get the correct one from Loni and update the example. It should work then.

13:54 <gnikunj[m]> hkaiser: aah that's why

13:54 <hkaiser> I'll do the build system fixes

13:55 <gnikunj[m]> yes, I'll update the 1d stencil codes so it doesn't seg fault

13:55 <hkaiser> thanks

14:06 <gnikunj[m]> hkaiser: what is this parameter: HPX_WITH_ASYNC_MPI?

14:07 <hkaiser> gnikunj[m]: that enables the async mpi module (executor)

14:07 <gnikunj[m]> does it mean that async will use MPI for actions if it is turned on? If yes, then why do we have HPX_WITH_PARCELPORT_MPI?

14:07 <hkaiser> no, that's unrelated

14:07 <gnikunj[m]> aah got it

14:30 <gnikunj[m]> hkaiser: are we meeting rn?

14:32 <hkaiser> gnikunj[m]: yes, sec

18:53 <bita> hkaiser, do we have a meeting at 1:00? I will present those materials at 4:00

20:15 <hkaiser> diehlpk_work: can we skip our meeting today?

20:18 <diehlpk_work> hkaiser, 10 minutes would be great

20:18 <diehlpk_work> SO many things are goind on

20:18 <hkaiser> ok

20:23 <diehlpk_work> hkaiser, We can start now, if you like

20:24 <hkaiser> ok, sec

20:25 <hkaiser> what link do you use?

20:25 <hkaiser> diehlpk_work: ^^ ?

20:26 <diehlpk_work> 334 041 0194

20:49 <gnikunj[m]> hkaiser: figured out what's causing it. Can't figure out why the current one doesn't work though. https://github.com/STEllAR-GROUP/hpx/blob/46cd4811dba4102ce9c482ea85de09d2809938a5/libs/full/resiliency_distributed/tests/performance/replay/1d_stencil_replay_distributed.cpp#L399

20:51 <gnikunj[m]> replacing that with std::vector<hpx::id_type>(1, hpx::find_here()) makes the code work just fine. Initially I thought it could be that locality.size() is less than the value of counter that's passed to operator[] but that's not the case either. It's a strange behavior and I'll see to it tomorrow morning.

21:11 <weilewei> Oh, freenode IRC now has a small overview windows for the posted links now

21:11 <weilewei> *preview window