hkaiser changed the topic of #ste||ar to: STE||AR: Systems Technology, Emergent Parallelism, and Algorithm Research | stellar-group.org | HPX: A cure for performance impaired parallel applications | github.com/STEllAR-GROUP/hpx | This channel is logged: irclog.cct.lsu.edu | Everybody: please respond to the documentation survey: https://forms.gle/aCpNkhjSfW55isXGA
hkaiser has quit [Quit: Bye!]
K-ballo has quit [Quit: K-ballo]
<rachitt_shah[m]> <ms[m]> "are you comfortable with..." <- I'm good, I would need help later with a sphinx build without API docs. I removed breathe from the extensions, but still APIs are rendering
<rachitt_shah[m]> Also working on the deployment, might need a hand with that
<rachitt_shah[m]> I'm playing with the conf.py.in, hoping it works
K-ballo has joined #ste||ar
hkaiser has joined #ste||ar
<K-ballo> > "-pthread" not available for language CXX.
<K-ballo> that's odd, we shouldn't be setting -pthread manually
<K-ballo> THREADS_PREFER_PTHREAD_FLAG should be set for FindThreads
<hkaiser> K-ballo: could it be that our bind implementation is not very SFINAE-friendly?
<K-ballo> define "very"
<hkaiser> I running into issues, for instance: https://gist.github.com/hkaiser/6ebbb454fcc711fcd50223b73e584968, which fails compiling
<K-ballo> it's impossible to be entirely SFINAE-friendly until deducing-this
<hkaiser> well, not at all in my case
<hkaiser> I don't think it's caused by the non-deducable this
<K-ballo> no, that one should work
<hkaiser> I thought so, but it doesn't
<K-ballo> what's the compiler output? I'm not in a condition to try locally
<hkaiser> sec
<hkaiser> K-ballo: added it as a comment to the gist above
<hkaiser> it instantiates member_type<0> for the first overload (which doesn't take any arguments)
<K-ballo> yeah, there's no bounds check on placeholders, mmh
<K-ballo> looks like I broke it when I introduced member_pack
<hkaiser> should I create a ticket for you?
<K-ballo> we don't want member_pack's get to sfinae, but bind was previously implicitly depending on util::get sfinaeing
<hkaiser> yes
<K-ballo> sure, but it may take a while for me to get to it
<hkaiser> K-ballo: any hint how to approach a fix would help as well
<K-ballo> bind_eval_placeholder, check if I is in bounds for US
<K-ballo> template <typename T, typename... Us, typename Enable = std::enable_if_v<(I < sizeof...(Us))>
<K-ballo> static constexpr HPX_HOST_DEVICE decltype(auto) call(
<hkaiser> ok
<hkaiser> K-ballo: that pushes the error further up, I'll have a look
<K-ballo> a =deleted counterpart overload would produce orders of magnitude better errors when calling bound with insufficient arguments
<K-ballo> looks like invoke_bound_result isn't sfinae-friendly at all
<K-ballo> so I didn't broke nothing :)
<K-ballo> curious... I even wrote a paper on the sfinae-ness of std::bind
<hkaiser> K-ballo: see #5489 for a possible solution - not the nicest one, but seems to work
ct-clmsn has joined #ste||ar
ct-clmsn has quit [Ping timeout: 250 seconds]