04:14 <hkaiser> Yorlik: please see #3727 for the example you requested

14:56 <Yorlik> o/

14:56 <Yorlik> hklaiser - I looked at the PR

14:57 <Yorlik> hkaiser: Looks all nice and good to me. So - will I have patch hpx to make this work?

14:57 <Yorlik> Or wait for change?

14:58 <hkaiser> Yorlik: if somebody approves the PR it will be in HPX ;-)

14:58 <Yorlik> Allright - I'l just wait for that. I have to do some other things beforehand anyways.

14:58 <hkaiser> Yorlik: a comment on the PR from you might help the process ...

14:58 <Yorlik> Thanks a lot for bringing this in - saves a ton of problems with one simple move.

14:59 <hkaiser> well, the allocator is everything but simple, but the rest is so, indeed

14:59 <Yorlik> I ciouzld ask for someone to please review it :)

14:59 <hkaiser> sure, every bit helps

14:59 <Yorlik> Yes - the allocator is a work on its own - but now we have an iterface

15:00 <Yorlik> Though the allocator is the biggest portionj of it, the small cghanges are the door openers here.

15:01 <Yorlik> And since I never wrote an allocator in practise it's big help

19:27 <Yorlik> Is there any way in this construction to guarantee, that the order of component assignement is always correct, like component_1 first? I tried constructiong a static_assert, but the expression never evaluated to a constant.

19:27 <Yorlik> class composit : public Interface, public component_1<Interface>, public component_2<Interface>{};

20:08 <hkaiser> Yorlik: base classes are initialized in the sequence they are derived from

20:10 <Yorlik> The problem is checking a static that is not const doen't give me a const for the static_seert.

20:10 <Yorlik> I tried dropping a flag in a static in the Interface and chacking against that flag

20:11 <Yorlik> but it didn't work - the compiler, ofc could not know, that I only change thios in the constructor

20:12 <hkaiser> Yorlik: I need to have more context to answer

20:12 <hkaiser> heller_: interesting read: https://link.springer.com/article/10.1007/s10766-018-0561-2

20:12 <Yorlik> So - though the static uint acting as a bitfield of which inheritence has already taken place effectivley acts like a const after instantiation, this in not something the comopiuler can know about and I don't have a way to get that information about the order of initialization into a compile time assert

20:12 <Yorlik> Lemme slap together a gist really quick

20:19 <Yorlik> hkaiser: https://gist.github.com/McKillroy/96645858b5f878bc11e08eae4dc315ac

21:03 <hkaiser> Yorlik: make your is constexpr

21:04 <Yorlik> The id ?

21:04 <hkaiser> yah, and the flags

21:13 <Yorlik> But I need to change the flags of derived when adding the components

21:13 <Yorlik> Making everyhing const violates that.

21:14 <Yorlik> I have a weird feeling I'll need some strange metaprogramming to achieve enforcing the order constraints

21:43 <hkaiser> Yorlik: I don't think so, the structure of your entity type is copletely compile-time defined, so it the value of your flag

21:44 <Yorlik> Every constructor of a component adds a flag to the CRTPed entity Interface

21:44 <Yorlik> That violates the constness

21:45 <Yorlik> The compiler doesn't understand it's meant as an "assembled const" if you could call it so

21:47 <hkaiser> Yorlik: but the structure is compile-time defined, so you should be able to assemble the flags at compile time too

21:48 <Yorlik> I am getting errors, when trying to or a flag into the static const mask of the entity Interface

23:26 <hkaiser> Yorlik: uhh, why's that?

23:26 <hkaiser> I can show you how to statically initialize the mask

23:29 <Yorlik> Do you have a working example?

23:30 <Yorlik> I couldn't figure it out so far

23:30 <Yorlik> I ended up putting the multiple inheritance into a typename ...Ts list

23:31 <Yorlik> Still have no solution - playing around learning .. but still stuck

23:31 <Yorlik> I'm still using CRTP to initialize the entity mask

23:32 <Yorlik> And I'm doing it in the constructors of the components

23:33 <Yorlik> I made a template function to have a static assertion from comparing two coimponent types, but thats not yet the solutiion

23:33 <Yorlik> like: assert_is_smaller< physics<IEntity>, container<IEntity> >

23:36 <Yorlik> I think the constructor probably is the wrong place because all the static consts are done then already

23:38 <Yorlik> However - it's getting really later here and I direly need sleep.

23:39 <hkaiser> Yorlik: https://gist.github.com/hkaiser/552d7d30b45062d5ea3aadd1d79fbb89