00:02 <gonidelis> Perfect. Thank you! Do you think forming a proposal including both range based algorithms and missing parallel algorithms would be too much? On the other hand which of the above do you think could be implemented the next few weeks in order for me to provide you with a sample work? (ps: is anyone else working on any of the above?)

00:31 <hkaiser_> gonidelis: nobody is working on those currently

00:31 <hkaiser_> I'd start with simple ones if you want to give a sample (not sort related)

00:32 <hkaiser_> the range stuff or the end iterator changes are easy enough

18:59 <gonidelis> Should I create a *component* for my custom app is it necessary to implement it in a way that it is represented by a *server class* and a *client class* according to your accumulator example? Should I use the accumulator template in order to create my own component? Is that a standard HPX architecture?

19:06 <zao> If you've got a need to invoke things on remote localities, a component is probably the thing you want.

19:06 <zao> (I'm learning a lot about HPX from you, I've very rarely actually used the thing, I just build it :D)

19:08 <gonidelis> Haha... Yeah that's the point actually. AIthough I think that I need to play around with it (maybe create some test-app examples) in order for me to be able to contribute with the building team

19:09 <gonidelis> As for the remote localities: isn't that what an action is about?

19:11 <gonidelis> Maybe because components expose methods as actions, we should deploy actions through the use of components... (only?) maybe...

19:19 <zao> I guess components are to classes what actions are to functions?

19:20 <gonidelis> Yeah that analogy seems sensible to me...thank you

19:21 <zao> hpx::cout is a decent concrete example I think - a proxy out on localities that forward to the instance on the console locality.

19:34 <Yorlik> hkaiser: yt?

19:35 <hkaiser> Yorlik: here

19:35 <Yorlik> I think I need a little bit of help with this one: https://hastebin.com/defeqitabi.cpp

19:35 <Yorlik> Trying to make a member metthod a template action

19:36 <Yorlik> Also I wonder what would I have to do about stuff that usually goes into the .cpp?

19:36 <hkaiser> should be &gameobject::send_message<M_>, note, don't use the same template argument name

19:37 <hkaiser> or sorry, forget what I said, simply use &gameobject::send_message<M>

19:37 <hkaiser> or &gameobject::template send_message<M>

19:37 <Yorlik> OK - anything I have to put in the cpp?

19:37 <hkaiser> no

19:37 <Yorlik> The gameobject is NOT a template

19:37 <hkaiser> sure

19:37 <hkaiser> otherwise it would be &gameobject<something>::send_message<M>

19:38 <Yorlik> OK

19:38 <Yorlik> I'll try ...

19:52 <Yorlik> hkaiser: Something is still wrong: https://hastebin.com/atekoceyoz.cpp

19:53 <Yorlik> forget it - the T...

20:18 <Yorlik> hkaiser: It still explodes - even after I cleaned up all possible junk getting in the way: https://hastebin.com/inufadosol.cpp Ideas?

20:19 <Yorlik> So - no old macros lingering or anything

20:20 <hkaiser> it's the 'uint8_t ( send_message_action::* )( M )' it should use the name of your class, most probably uint8_t ( game_object::* )( M )

20:24 <Yorlik> hkaiser: Would you consider this correct? (At least it compiles) https://hastebin.com/ucujekuget.cpp I think there now is a documntation demand for "hpx::actions::make_action<>" :)

20:24 <hkaiser> looks ok

20:25 <hkaiser> and yes for the docs

20:25 <Yorlik> Allright - time to test some more - wnat me to make an issue for the doc?

20:43 <Yorlik> hkaiser: It seems to work, just Lua explodes somewhere - probably I need to lock sth. However - the basic sending through AGAS seems to work.

20:43 <hkaiser> nice

21:29 <gonidelis> day #3 of familirizing myself with HPX and I must say that it is one of the most interesting libraries that I have encountered! Just finished with your quick-guide examples and I have watched Mr.Kaiser's talk on CppCon 2016. The talk was just on point and you could not imagine how much did it help in order to wrap up the concepts that I have

21:29 <gonidelis> learned so far. Great work! The next one or two days I am planning on working with your tutorials repo in order to gain more experience on working with HPX and I really hope that by the end of this studying I will be able to implement a simple matrix multiplication with HPX (as you propose on your GSoC page).

21:29 <gonidelis> Before I get further let me address some questions to the community though: The last three years I have been programming mainly on python and C(quite proficient) on various projects. Trying to be honest with you I have some good experience with C++ but not in the same scale as I have with C. I certainly know the general theory and have good

21:29 <gonidelis> undestanding of the major C++ concepts but I have not implemented any large-scale projects on C++ so far (just various medium-scale scripts). Reading your src code is not that easy but after spending let's say a couple of hours reading a module at the end I can totally get it but certainly not write it from scratch. Another example of my C++ level

21:29 <gonidelis> of skills is that it does not seem that easy to me the way you organize your code (hpp and cpp files) or your namespaces. Creating a Cmake file in order to build that module is more of a challenge. These are some random but characteristic examples of my C++/HPX level of expertise.

21:29 <gonidelis> According to that, do you think that trying to read and understand or even write some HPX code within the next 2 weeks would be fissible for a programmer with my kind of skills? If not I would still love to keep working on the notions mentioned above in order to be a future contributor on your HPX project though regardless of GSoC. I would really

21:29 <gonidelis> appreciate an honest answer as far as realism is concerned. Thank you all!ps: I have worked with parallel and destributed systems quite a lot the past six months and I must say the parallelization concepts that you expose with HPX are quite comprehensible to me.

21:38 <zao> It's ugly inside so that the outside is nicer to use :P

21:43 <zao> I'd reckon that implementing parallel algorithms on HPX would be more about the algorithmic stuff and less about any icky internals. You seem to pick things up well enough, the build and layout will probably fall into place eventually.

21:44 <zao> (I'm not involved in GSoC or active HPX dev, so take the things I say with a fistful of salt :D )

21:46 <hkaiser> gonidelis: need to run now, will respond later

21:47 <gonidelis> is there anyway for me to receive your messages in case i am offline?

21:47 <zao> There's a link in the topic to see a log.

21:48 <zao> Click that and #ste||ar in the left sidebar.

21:51 <gonidelis> oh! great. thank you very much