K-ballo changed the topic of #ste||ar to: STE||AR: Systems Technology, Emergent Parallelism, and Algorithm Research | stellar.cct.lsu.edu | HPX: A cure for performance impaired parallel applications | github.com/STEllAR-GROUP/hpx | Buildbot: http://rostam.cct.lsu.edu/ | Log: http://irclog.cct.lsu.edu/
<K-ballo> master: 1.47 GB (1,587,318,094 bytes)
<K-ballo> object-libraries: 1.15 GB (1,240,473,946 bytes)
nanmiao11 has quit [Remote host closed the connection]
nanmiao11 has joined #ste||ar
<hkaiser> K-ballo: why is that?
<K-ballo> separate libraries are bound to duplicate a lot of stuff
hkaiser has quit [Quit: bye]
bita has quit [Ping timeout: 240 seconds]
akheir1 has quit [Quit: Leaving]
RostamLog has joined #ste||ar
bita has quit [Read error: Connection reset by peer]
k-ballo[m] has quit [Quit: Idle for 30+ days]
wladimir[m] has joined #ste||ar
k-ballo[m] has joined #ste||ar
hkaiser has joined #ste||ar
<Yorlik> Just tried a clean build of 1.5.0 and am getting this error: "hpx\runtime\parcelset\encode_parcels.hpp(220): error C2027: use of undefined type 'hpx::actions::base_action'"
<hkaiser> Yorlik: just building HPX?
<Yorlik> Yes
<hkaiser> grrr
<hkaiser> what's different to our builders, then?
<hkaiser> what platform, what cmake options?
<Yorlik> Lemme check some things ...
<Yorlik> Boost 1.73 (built w. vcpkg), mimalloc 1.64 (also vcpkg), hwloc 1.11.7-4 (vcpkg), release build (didn't check debug yet)
<Yorlik> CMake switches:
<Yorlik> -DCMAKE_TOOLCHAIN_FILE=${CMAKE_TOOLCHAIN_FILE} -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE} -DCMAKE_INSTALL_PREFIX=${HPX_INSTALL_DIR}
<Yorlik> -DCMAKE_VERBOSE_MAKEFILE=${CMAKE_VERBOSE_MAKEFILE} -DHWLOC_ROOT=${HWLOC_ROOT} -DHPX_WITH_CXX17=ON
<Yorlik> -DHPX_WITH_MALLOC=mimalloc -DHPX_FILESYSTEM_WITH_BOOST_FILESYSTEM_COMPATIBILITY=ON -DBUILD_SHARED_LIBS=${HPX_BUILD_SHARED_LIBS}
<Yorlik> -DHPX_TESTING_WITH_TESTS=${HPX_ALL_TESTS} -DHPX_WITH_BACKGROUND_THREAD_COUNTERS=ON -DHPX_WITH_COROUTINE_COUNTERS=ON
<Yorlik> -DHPX_WITH_EXAMPLES=OFF -DHPX_WITH_IO_COUNTERS=OFF -DHPX_WITH_PARCELPORT_ACTION_COUNTERS=ON
<Yorlik> -DHPX_WITH_THREAD_CUMULATIVE_COUNTS=ON -DHPX_WITH_THREAD_IDLE_RATES=ON
<Yorlik> Platform is Windows 10, Latest MSVC
<Yorlik> Hpx is built with CMake as "Open Folder" project
<Yorlik> hkaiser: ^^
<Yorlik> hkaiser: This gist has the two error causing parts of the log:
<Yorlik> hkaiser: I switched to stable branch and removed a ton of options, including mimalloc. I'm still getting errors. I added a file with these to the gist: https://gist.github.com/McKillroy/2a289e24849a3209ffd66482ce359806
shahrzad has joined #ste||ar
<hkaiser> Yorlik: yah, that one was fixed this morning
<hkaiser> missing #include, no idea how that slipped through
<K-ballo> we must not be building for 17 on windows
<K-ballo> I only ever build the default
parsa has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
parsa has joined #ste||ar
weilewei has joined #ste||ar
<hkaiser> K-ballo: I do build using c++17, so no idea how that could happen
akheir has joined #ste||ar
hkaiser has quit [Read error: Connection reset by peer]
hkaiser has joined #ste||ar
<K-ballo> hkaiser: are we setting the right flags?
<hkaiser> not sure
<Yorlik> hkaiser: With the last commit it's now building with all our set options. Not sure what this 1.5.0 issue was - seems you fixed some issues in the meantime already. Maybe it might be time for a 1.5.1 that is bugfixed soonish?
<hkaiser> Yorlik: this is the first issue reported with 1.5
<hkaiser> but sure, let's consider it
<Yorlik> I had issues building 1.5.0 too
<hkaiser> ms[m], rori: pls see ^^
<Yorlik> the first file in the gist has the log. The second was the latest stable. With head now it compiled for me.
<hkaiser> Yorlik: yes, the second issue was caused by that missing #include :/
<hkaiser> the first - I have no idea
<Yorlik> It was a clean build - however - I'd never exclude issues in the depths of our way to build it.
<Yorlik> I only don't understand how the last failed build could become stable tagghed.
<Yorlik> Is windows still not built for stable testing?
<hkaiser> Yorlik: happens in c++17 on MSVC only
<hkaiser> stable tag depends only on the CirclCI builder
<Yorlik> Is that not tested for stable? If not I'd like to suggest that.
<hkaiser> Yorlik: sure, would be nice to include other builders into the decision whether to tag things as stable
<Yorlik> But maybe it's again a resource issue ...
<hkaiser> I don't know however, how this could be achieved
<ms[m]> hkaiser: thanks for the heads up
<ms[m]> Do you know if the windows builder was broken in the meantime? (If not, making stable depend on it would be of little use)
<ms[m]> We do have a few windows builders so we can make one of them do 17 at least
<ms[m]> And yes, if we're doing 1.5.1 it's about the right time, wouldn't be a big deal at all
<ms[m]> 4946 is another potential candidate for 1.5.1, completely non-breaking
<ms[m]> the new algorithms ranges prs could also maybe go in 1.5.1, but don't have to
<zao> std:c++latest or bust :P
bita has joined #ste||ar
weilewei has quit [Remote host closed the connection]
<K-ballo> -Zc:preprocessor
<K-ballo> -permissive-
nanmiao11 has quit [Remote host closed the connection]
hkaiser has quit [Read error: Connection reset by peer]
hkaiser has joined #ste||ar
diehlpk_work has quit [Remote host closed the connection]
diehlpk_work has joined #ste||ar
<khuck> whoever might know... can HPX be built with sanitizer flags (-fsanitize=memory) and if so, are there special flags to do that? I am trying to hunt down a heisenbug
<hkaiser> khuck: there is a cmake option for that
<ms[m]> khuck: -DHPX_WITH_SANITIZERS=ON, but you still have to set your CXXFLAGS
<ms[m]> you may also get some results from hpx itself...
<ms[m]> but maybe that's what you're looking for
<hkaiser> also, it crashes when used with clang iirc
<hkaiser> gcc should work
<K-ballo> cleaner debug/print.hpp parsing time down to 58%
<khuck> cool, thanks. So I need to set both -DHPX_WITH_SANITIZERS=ON *and* CXXFLAGS="-fsanitize=address" (and probably CFLAGS?)
<khuck> and LDFLAGS?
<ms[m]> khuck: actually...
<ms[m]> LDFLAGS as well, and -fno-omit-frame-pointer on CXXFLAGS
<khuck> oh man
<khuck> turtles, all the way down...
<ms[m]> K-ballo: nice!
<ms[m]> we could consider adding those automatically...
<khuck> ms[m]: I assume that after I build HPX, I will need to use the same flags with Phylanx (which uses HPX)?
<ms[m]> hmm, yes but I think they should propagate to dependent projects
<ms[m]> unsure now... in any case, if they're not propagated you will need to propagate them
<khuck> thanks
jaafar has quit [Ping timeout: 260 seconds]
jaafar has joined #ste||ar
<hkaiser> K-ballo: nice!
k-ballo[m]1 has joined #ste||ar
mathegenie[m]1 has joined #ste||ar
k-ballo[m] has quit [Ping timeout: 246 seconds]
mathegenie[m] has quit [Ping timeout: 246 seconds]