15:00 <diehlpk_work> ms[m], Just to confirm you will take care of the GSoD application?

15:45 <ms[m]> diehlpk_work: sure, I'll give it a go

17:57 <jedi18[m]1> How do I run the tests on visual studio? I right click on adjacentfind_test and click on "Run Tests" but that dosen't really do anything :/

17:58 <gonidelis[m]> you run `ctest -R <target>`

17:58 <gonidelis[m]> like `ctest -R tests.unit.modules.algorithms.adjacentfind`

17:59 <gonidelis[m]> use auto complete with tab. it will help ;)

17:59 <hkaiser> jedi18[m]1: right click on the test, 'Set as startup project', and run as usual

17:59 <hkaiser> (F5)

17:59 <gonidelis[m]> hkaiser: 0.0

18:04 <jedi18[m]1> Thank you! Where do I place the boost dll files? Build/Debug/bin?

18:07 <jedi18[m]1> hkaiser: it just prints "using seed: 1230896067"

18:10 <hkaiser> so it runs

18:10 <hkaiser> it would complain loudly if something was wrong

18:11 <jedi18[m]1> Ohh ok thanks. No I stashed the changes and tried running it first, I'll pop those changes and see if the tests run

18:13 <jedi18[m]1> Ok yeah the tests run, I just changed the tag_fallback to tag_fallback_invoke and hpx::functional::tag<adjacent_find_t> to hpx::functional::tag_fallback<adjacent_find_t>. Is that really all that was required to be changed?

18:14 <gonidelis[m]> jedi18: yes, afaik

18:14 <gonidelis[m]> compiler and CI will be the final judges though

18:15 <jedi18[m]1> So do I create a PR then?

18:20 <jedi18[m]1> BTW are there some other things I'd need to learn not mentioned in the C++ series? Like concepts.

18:51 <gonidelis[m]> jedi18: reading TSs I in general I would say, gettin close to the standardization process ;)

18:51 <gonidelis[m]> not just concepts

18:52 <gonidelis[m]> btw we do not use concepts (yet)

19:06 <hkaiser> jedi18[m]1: yes, pls create a PR

19:06 <pedro_barbosa[m]> Hey guys, I was adapting one of the exercises from the tutorials (original exercise: https://github.com/STEllAR-GROUP/tutorials/blob/master/examples/00_exercises/solution9.cu) and ended up with this https://gist.github.com/PADBarbosa/ebdd78dae9a8b33ea93118953c4288e2, but for some reason when I run it the vh array is filled with 2s instead of 4s, anyone knows what I'm doing wrong?

19:14 <K-ballo> now if only I could raise a warning/error whenever sfinae happens "accidentally" ...

19:17 <ms[m]> pedro_barbosa: most likely you'll need to make sure that the for_each happens before the copy, as it is there's nothing to tell the runtime that should happen

19:18 <ms[m]> either with an explicit f.get(), or better by sticking the copy in a continuation with f.then()

19:18 <ms[m]> note that if you use hpx::execution::par(hpx::execution::task) instead of just hpx::execution::par you get a future back from for_each and you don't have to wrap it in async

19:19 <pedro_barbosa[m]> Going to give that a try, thanks

19:21 <ms[m]> K-ballo: what sort of "accidental"? :P

19:22 <K-ballo> pretty much anything that isn't an enable_if, essentially.. except there's no point for enable_if either with requires

19:22 <K-ballo> maaaaaaaaaaaaybe some invoke_result uses would be reasonable, but I'm not entirely convinced

19:22 <K-ballo> better to be explicit

19:27 <gonidelis[m]> lol

19:29 <pedro_barbosa[m]> <ms[m] "either with an explicit f.get(),"> Tried doing a then but was having some truble so in order to test if that was the problem I used a f.get() but the problem remains, the final array is filled with 2s instead of 4s

19:31 <ms[m]> pedro_barbosa: v is passed by value to fun1?

19:31 <ms[m]> I think that makes a copy, assuming it behaves the same as std::vector (I haven't used those much)

19:42 <pedro_barbosa[m]> yeah that was the problem, was able to fix the problem

19:42 <pedro_barbosa[m]> * yeah that was the problem, was able to fix it

19:42 <pedro_barbosa[m]> * yeah that was the problem, was able to fix it, thanks!

19:48 <jedi18[m]1> <hkaiser "jedi18: yes, pls create a PR"> Oh ok, I did :)

19:57 <hkaiser> jedi18[m]1: thanks!

20:04 <jedi18[m]1> hkaiser: I dind't get your comment, add the changes to which file?

20:05 <hkaiser> to the related container_algorithm

20:05 <jedi18[m]1> <hkaiser "to the related container_algorit"> Oh right ok sure

20:12 <hkaiser> jedi18[m]1: I added the link now, sorry

20:17 <jedi18[m]1> <hkaiser "jedi18: I added the link now, so"> np, thanks for reviewing the PR!

21:21 <fodinabor> hey :) I do get a "negative entry in reference count table" exception with HPX 1.5.1 (only when running on multiple nodes) and am wondering how to go about debugging this

21:39 <heller> Hey. That sounds bad. Is it easily reproducible?

21:41 <heller> Like in every run? If yes, then the best thing would be to figure out when this is happening. This could be done with running the application with --hpx:attach-debugger=exception

21:41 <heller> And then attach to the pid and inspect the threads

21:42 <hkaiser> uhh ohh, we haven't seen one of those in a loooong time

21:46 <fodinabor> Uhm... well it was pretty consistent the last few hours.. but not in the last 10 runs..

21:48 <fodinabor> what would I see with the debugger attached? the decrement_sweep function is not called, when the reference is decremented directly or is it? Or is there a way to determine what kind of object seems to be the problem from the GID, which is printed in the stack trace? raw({000000030000004e, 0000000001d66c60})

21:51 <heller> The idea is to figure out which operation triggers the exception

21:52 <heller> Debugging those errors is hard... The best thing would be a small self contained example to reproduce the issue

21:53 <fodinabor> given that it ran stable the last x tests (without changing too much).. any chance this could be caused by an instable network or similar?

21:54 <heller> Unstable network is highly unlikely

21:54 <fodinabor> okay.. will try to reproduce.. and have a look with the debugger

21:54 <heller> If you changed not too much, you changed something. That might have made the difference already

21:55 <heller> What we need is at least code snippets which you think are causing the problem

22:00 <heller> Or something along those lines

22:01 <fodinabor> what counts as small example? :P we're basically running an extended version of the stencil-1d 8 example, extended to do some version of 3d stencil... and already tried to stabilize some race-conditions from that example.

23:34 <fodinabor> Fixed another race condition there (got promise overwrite errors in the meantime).. seems to be rather stable for now..

23:34 <fodinabor> will check tomorrow, whether it still works ;-)