00:25 <bita_> this https://github.com/microsoft/vcpkg/pull/5064 solved my vcpkg problem

00:57 <bita_> I don't get vcpkg errors anymore, but unfortunately, I still get these errors: https://gist.github.com/taless474/671a361724f1acf8aae85abfa086d8fa

12:42 <nikunj97> K-ballo, if I have to suggest books to C++ novice to learn modern C++ and advanced C++ features (template metaprogramming, SFINAE etc.), which one's should I suggest? Also in what order?

12:47 <K-ballo> nikunj97: I wouldn't know

12:48 <nikunj97> K-ballo, yea. That's what I said as well. Wanted to know if there's actually a guide or so for it.

12:48 <K-ballo> there's a popular SO question on c++ books

12:48 <K-ballo> https://stackoverflow.com/questions/388242/the-definitive-c-book-guide-and-list

12:49 <nikunj97> aah, thanks!

12:50 <zao> Not sure about the quality of them, but the C++ Weekly videos on particular parts of the language are probably decent introductions to a lot of things.

12:51 <nikunj97> zao, talking about Jason Turner?

12:51 <zao> yup

12:51 <nikunj97> true, I watch them regularly.

13:29 <nikunj97> hkaiser, yt?

13:29 <hkaiser> nikunj97: in a meeting

13:29 <nikunj97> ohh ok.

15:02 <nikunj97> hkaiser, https://gist.github.com/NK-Nikunj/527b0bcd36d00032eb02c972346039b4

15:03 <nikunj97> hkaiser, this is the error: https://gist.github.com/NK-Nikunj/527b0bcd36d00032eb02c972346039b4#file-error-log

15:04 <hkaiser> nikunj97: I think you can write async(foo, binpacked(localities), ...), but I have no idea why it doesn't compile, will look

15:04 <hkaiser> no need to create an instance yourself

15:04 <hkaiser> https://github.com/STEllAR-GROUP/hpx/blob/master/hpx/runtime/components/binpacking_distribution_policy.hpp#L364

15:05 <nikunj97> yea, I tried binpacked first. But it didn't work. I thought, creating an instance may mitigate the issue.

15:05 <hkaiser> nikunj97: let me have a look

15:05 <nikunj97> sure! Btw do we have a node level load balancer?

15:06 <hkaiser> load balancing in between nodes?

15:06 <nikunj97> yes

15:06 <hkaiser> parsa is workin gon it as we speak

15:10 <hkaiser> nikunj97: the compiler clearly tells you what the problem is (at least clang does)

15:11 <Nikunj__> should use clang next time :/

15:11 <Nikunj__> hkaiser, what is the error exactly?

15:12 <hkaiser> hpx/async_distributed/async.hpp(247): message: candidate template ignored: substitution failure [with Component = hpx::actions::detail::plain_function, Signature = int (), Derived = answer_action, DistPolicy = hpx::components::binpacking_distribution_policy, Ts = <>]: no member named 'async_result' in 'hpx::components::binpacking_distribution_policy'

15:12 <hkaiser> Nikunj__: the problem is that we screwed up here ;-)

15:13 <hkaiser> distribution policies are somehow two in one: distributed allocation and distributed invocation

15:13 <hkaiser> but the binpacking policy never got the distributed invocation part implemented

15:13 <hkaiser> but that could be eaily added, I think

15:13 <Nikunj__> so we need to implement the distributed invocation?

15:13 <Nikunj__> that's why colocated worked and distribution didn't?

15:14 <hkaiser> something like https://github.com/STEllAR-GROUP/hpx/blob/master/hpx/runtime/components/colocating_distribution_policy.hpp#L146-L249

15:14 <nikunj97> ohh so apply and async functions for them

15:14 <nikunj97> I get it

15:15 <hkaiser> nikunj97: the difference is the id_type, it's 'delivered' by the policy

15:16 <nikunj97> as in?

15:17 <hkaiser_> nikunj97: I was never happy with the way things were implemented

15:17 <nikunj97> you mean binpacking or all distributed policies?

15:17 <hkaiser_> the whole execution could be separated from the distribution policy as all itwould need would be to call get_next_target() on it, the rest should be the same for all

15:18 <nikunj97> true, there's redundancy in the code

15:19 <hkaiser_> nikunj97: this: https://github.com/STEllAR-GROUP/hpx/blob/master/libs/async_distributed/include/hpx/async_distributed/async.hpp#L108-L116

15:19 <hkaiser_> would have to be rewritten to just call get_next_target on the policy and dispatch to the base async() implementation

15:19 <nikunj97> aha, so it couldn't find the async function when called from here

15:19 <hkaiser_> yes

15:20 <hkaiser_> then we could remove the execution part from all distribution policies

15:20 <nikunj97> this does not sound like a great implementation

15:20 <hkaiser_> let me do that, shouldn't take too long

15:20 <nikunj97> I think, this execution policies should just be wrappers to how it will be executing and not include the invocation themselves. That should be a template specialization within async

15:21 <nikunj97> I think I read something similar in Andrei's book

15:21 <hkaiser_> well, that's what this (https://github.com/STEllAR-GROUP/hpx/blob/master/libs/async_distributed/include/hpx/async_distributed/async.hpp#L108-L116) is all about, no?

15:22 <hkaiser_> I think what needs to be done, the distributon policies shouldn't have any code, all they need is the get_next_target function

15:23 <hkaiser_> not sure why I have not seen this before :/

15:23 <nikunj97> well technically yes. But having invocation within all distribution policies sound a bit redundant. The policies themselves assign the localities on which they'll be invoked. One can use a single invocation and the policy to make it work, no?

15:42 <hkaiser_> nikunj97: that's what I meant

15:42 <hkaiser_> also, I think it makes sense for the binbacking policy not to do execution

15:43 <nikunj97> ohh I must've misinterpreted then

15:43 <nikunj97> hkaiser_, yes agreed

15:43 <hkaiser_> this one is all about allocation anyways, but I guess for some work-load balancing it might be useful

15:43 <hkaiser_> nikunj97: I'll work on some changes for the policies

15:44 <nikunj97> I need to do a call with parsa. I need load balancing capabilities to have a working distributed resiliency model

15:44 <nikunj97> hkaiser_, thanks a lot

18:25 <nanmiao11> I got an error "/home/nanmiao/Documents/project/dev/src/phylanx/src/execution_tree/primitives/assert_condition.cpp:10:10: fatal error: hpx/distributed/iostream.hpp: No such file or directory #include <hpx/distributed/iostream.hpp>". Any one has the same issue?

18:26 <hkaiser_> nanmiao11: do you have a recent hpx?

18:26 <nanmiao11> use this one https://github.com/STEllAR-GROUP/hpx/commit/75884de4450597bacb2e8db98da84174254ce5ba

18:27 <hkaiser_> you might have to add HPX::iostreams as a dependency to your project

18:28 <hkaiser_> we might have to fix Phylanx - however, I'm not seeing this problem

18:29 <nanmiao11> Will try to add it.

18:30 <weilewei> hkaiser_ meeting now