00:08 <hkaiser> K-ballo: master is hanging for you?

00:08 <hkaiser> uhh

00:14 <K-ballo> not exactly... I'm seeing hangs after applying a change that isn't hit

00:15 <K-ballo> and it's not consistent, comes and goes

00:15 <K-ballo> so I'm left to assume it must happen on master too.. but it's confusing

00:28 <parsa> daissgr: ping

00:31 <hkaiser> parsa: Gregor is definitely asleep

00:32 <parsa> hkaiser: he keeps going on and off… wasn't on during day… i can't tell anymore

00:36 <daissgr> nope! I have now arrived

00:36 <daissgr> sorry for the delay!

00:36 <daissgr> I am just now writing an answer to patricks email

00:45 <Yorlik> I think I am running into a similar issue with the use of macros like yesterday

00:45 <Yorlik> I define a class in a namespace and want to register it as a component

00:46 <Yorlik> It live in a dll

00:46 <Yorlik> I also want a method of this class to be registered as an action

00:47 <Yorlik> I put this macro into the source: using solstate = lua::solstate;

00:47 <Yorlik> using solstate_type = hpx::components::component<solstate>;

00:47 <Yorlik> HPX_REGISTER_COMPONENT ( solstate_type, solstate );

00:47 <Yorlik> But the type is not known, because it is in the source and not in the header

00:47 <Yorlik> The example in the docs doesn't mention any Macros in the header for this

00:48 <Yorlik> I was looking at this: https://stellar-group.github.io/hpx/docs/sphinx/latest/html/manual/writing_distributed_hpx_applications.html#defining-components

00:48 <Yorlik> The example starts with the action definition

00:49 <Yorlik> but nothing about the component itself is in the header

00:49 <Yorlik> I couldn't imagine how this would work - I think the documentation is missing something here.

00:52 <parsa> daissgr: do you have some time now or in the next few days to discuss what to do about the radiation kernel?

00:53 <Yorlik> Argh .. forgot HPX_COMPONENT_EXPORT

00:53 <Yorlik> OK - ditch that question ...

00:53 <daissgr> Ah! The next few days would be a lot better! I had some issues with my train connections today so I arrived a lot later than I originally anticipated - right now I am rather tired :)

00:54 <daissgr> parsa: when do you have time do this over the next few days?

00:54 <parsa> i'm pretty much available all the time… this is one of the things i have to get done before if i want to graduate :)

00:57 <daissgr> If you want we can do this tomorrow. As Dirk mentioned during the last conference call we are away from Sunday until Wednesday, so I am having a hard time planning anything for those days

00:57 <daissgr> so for me either tomorrow or after Wednesday would work

00:57 <parsa> tomorrow works… will you be on IRC?

01:00 <daissgr> sure I can do that! Let's say 10pm my time, which should translate to 3 pm CST?

01:00 <daissgr> oh hey, I just checked another run I started earlier. Apparentely our current GPU kernels are now working under power as well :)

01:01 <hkaiser> nice!

01:02 <parsa> daissgr: good… talk you then

01:02 <daissgr> hkaiser: One thing I have noticed: On PowerPC Octo-Tiger is crashing immediately when it is called to to many command line arguments before there is any output. I simply sidestepped this problem by using less arguments and moved my initialisation to a .ini file

01:03 <daissgr> could this be HPX related? I have never seen this behaviour before. Right now it is not critical, just annoying

01:03 <hkaiser> daissgr: heller_ fixed a stack overflow during ini file handling on Power today, not sure if that's related

01:04 <daissgr> parsa: Alright until tomorrow. That should give me time to over the radiation beforehand as well - right now I don't know to much about that part of the code

01:04 <hkaiser> see #3703

01:06 <daissgr> okay! Right now I am using a somewhat older HPX commit that Patrick recommended (65c22662ccd5c63f43421cf76ca29d8222bf7f23) that seems to work otherwise.

01:07 <daissgr> Once 3703 is merged I will try out HPX master again

01:09 <Yorlik> How would I invoke an action on a class? Just replace the location id with the instance id?

01:09 <Yorlik> like: hpx::apply<solstate_run_action> ( mSolId ); ?

01:09 <hkaiser> Yorlik: yes, exactly

01:10 <Yorlik> I still have some unknown typ problems, but I'll try to figure that out

01:10 <hkaiser> Yorlik: for the component you really can look at the accumulator example

01:10 <Yorlik> OK - I'll read that. Thanks !

01:15 <Yorlik> All these macros make my head spin and dizzy ;)

01:20 <Yorlik> is _action a default action identifier suffix if it is omitted in the macro?

01:20 <daissgr> excellent! Using multiple GPUs works as well. I'll just update my email from now then I can finally go to sleep :)

02:19 <Yorlik> What is this whole myType::wrapped_type::xyz thing in HPX_REGISTER_ACTION_DECLARATION ?

03:10 <hkaiser> Yorlik: let;s talk tomorrow, I'm off now

03:11 <Yorlik> Allright - Thanks !