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
K-ballo has quit [Quit: K-ballo]
diehlpk has joined #ste||ar
diehlpk has quit [Quit: Leaving.]
hkaiser has quit [Quit: Bye!]
tufei__ has quit [Quit: Leaving]
tufei has joined #ste||ar
tufei has quit [Remote host closed the connection]
tufei has joined #ste||ar
Yorlik has joined #ste||ar
Yorlik has quit [Ping timeout: 260 seconds]
Yorlik has joined #ste||ar
hkaiser has joined #ste||ar
<Yorlik> hkaiser: YT?
<hkaiser> here
<Yorlik> hkaiser oops - just didn't see it. You still there?
<hkaiser> Yorlik: here
<Yorlik> Allright
<Yorlik> I'd have time for that vcpkg thing if you want.
<hkaiser> Yorlik: thanks, can we defer that, please? I don't really have any additional cycles available these days :/
<Yorlik> NP. Just poke me whenever you want.
<hkaiser> wiil do, thanks
<Yorlik> Cheers :)
<hkaiser> ms[m]: does master configure properly for you?
<ms[m]> hkaiser: it does
<ms[m]> what do you see?
<hkaiser> CMake Error at cmake/HPX_Message.cmake:51 (message):
<hkaiser> MODULE_DEPENDENCIES for module hpx_mpi_base
<hkaiser> The module runtime_configuration should not be be listed in
<ms[m]> hrm, ok, you have that check enabled... let me see with that as well
<ms[m]> configures...
<ms[m]> that check is also a lie, runtime_configuration is definitely supposed to be there
<ms[m]> do you also get it with clean configuration?
<hkaiser> yes
<ms[m]> any special options that you're setting?
<hkaiser> hmmm
<hkaiser> does it work for you?
<ms[m]> yeah
<ms[m]> well, I should probably enable mpi to have the same environment as you :P
<hkaiser> the mpi_base module is listed as beeing part of full: https://github.com/STEllAR-GROUP/hpx/blob/master/libs/core/mpi_base/CMakeLists.txt#L33
<hkaiser> should be core
<ms[m]> yep, good catch
<hkaiser> orks with that change
<hkaiser> well, no quite
<hkaiser> hold on
<ms[m]> maybe?
<ms[m]> hkaiser: thanks! let's enable that check on one of the rostam builds as well (since they have mpi enabled)
<hkaiser> should I add this to the PR?
<ms[m]> should I push that or are you on it?
<ms[m]> either way
<hkaiser> n it
<hkaiser> on it
<ms[m]> we can actually enable it everywhere on ci
<hkaiser> I added it for gcc9
<hkaiser> but sure, I can add it to every config
<ms[m]> I'd enable it on all of them, there's no harm in that
<ms[m]> I can push...
<hkaiser> ok, pls do
<ms[m]> hkaiser: done
<ms[m]> thanks!
<hkaiser> let's merge this asap
<ms[m]> sure, please go ahead
<hkaiser> thanks
K-ballo has joined #ste||ar
<hkaiser> ms[m]: also, pls see https://github.com/STEllAR-GROUP/hpx/pull/5559
<ms[m]> hkaiser: thanks, approved
<hkaiser> thanks
K-ballo has quit [Ping timeout: 252 seconds]
diehlpk has joined #ste||ar
K-ballo has joined #ste||ar
diehlpk has quit [Client Quit]
Yorlik has quit [Ping timeout: 268 seconds]
<dkaratza[m]> I found a theme which requires a different version of sphinx, But if I install this sphinx version I get an error later on, so I guess its not compatible with the rest docs. Are interesting in changing the sphinx version as long as we decide we prefer this theme, or do you think it's not worth?
<dkaratza[m]> I found a theme which requires a different version of sphinx. But if I install this sphinx version I get an error later on, so I guess its not compatible with the rest docs. Are we interested in changing the sphinx version as long as we decide we prefer this theme, or do you think it's not worth?
<hkaiser> depends on how much effort it would be to make it work
<ms[m]> dkaratza: interested, but I'm not too optimistic at the moment about bumping those versions
<ms[m]> we'd likely need to do some debugging for the sphinx/breathe people to help them out
<ms[m]> if the theme is really nice I might try to find some time earlier for it ;)
<dkaratza[m]> <ms[m]> "if the theme is really nice I..." <- ok so maybe tomorrow i'll show you which themes i find good and we decide all together if sth is so nice that we would like wasting the time for
<hkaiser> +1
<hkaiser> ms[m]: we might have to split up the runtime_configuration modyule to separate the local stuff
<hkaiser> I'm seeing warnings here: https://github.com/STEllAR-GROUP/hpx/blob/master/libs/core/runtime_configuration/include/hpx/runtime_configuration/static_factory_data.hpp#L28-L35 complaining about inconsistent export/import modes as the definition of those functions is in the full module
<hkaiser> in runtime_distributed/runtime_support_server.cpp
<ms[m]> sigh...
<ms[m]> looks like those definitions could move to core though
<ms[m]> or do you know if they do something which would prevent that?
<hkaiser> it's related to the component stuff
<hkaiser> not sure if you want to have that in core
<hkaiser> the whole file belongs in full, but that will cause many other things to move
<ms[m]> There are plenty of other things that could be in full eventually but are still in core (e.g. plugins)
<ms[m]> If it's just those four functions and it's possible to move them on their own to core let's do that
<hkaiser> I'll investigate this
<hkaiser> fighting linker errors first...
diehlpk has joined #ste||ar
Yorlik has joined #ste||ar
diehlpk has quit [Client Quit]
nanmiao has joined #ste||ar
<nanmiao> Hi Dr. Kaiser, I have tested the pr, and I got compiling errors. I have sent the error info gist link to your email.
<nanmiao> hkaiser
<hkaiser> nanmiao: thanks
<hkaiser> ms[m]: I'm seeing a lot of duplicate symbol definition errors, how did this ever work on the windows CI builder :/
<hkaiser> nanmiao: does SHAD send const pointers?
<nanmiao> hkaiser Yes, see here, https://github.com/STEllAR-GROUP/SHAD/blob/master/include/shad/data_structures/local_set.h#L619 it sends const tuple, wherein the first and third elements are pointer.
<hkaiser> nanmiao: ok, that explains it
<hkaiser> I will fix the PR asap
<nanmiao> Thanks
<hkaiser> no, nvm
<hkaiser> need to expand the test to be sure
<nanmiao> ok
<hkaiser> we might need another overload for https://github.com/STEllAR-GROUP/hpx/pull/5555/files#diff-10764592844e9ec4e624e00b89eaf5a1bbe700eacaa27e189957683ce61d85ddR118, but for T const*, then the &p could be changed to const_cast<T*>(&p)
<hkaiser> 'll see what I can do
<K-ballo> const_cast!
<nanmiao> Thanks, I also tested another pr, https://github.com/STEllAR-GROUP/hpx/pull/5554, this one works properly. Thank you!
<hkaiser> K-ballo: the whole pointer serialization is bogus
diehlpk_work has joined #ste||ar
<diehlpk_work> hkaiser, ms[m] I could finally generate a correct cff file for hpx.
<diehlpk_work> I just need to add the badge to hpx's readme
<hkaiser> what badge?
<diehlpk_work> I know we have these commands |zenodo_doi
<diehlpk_work> I just do not know how to add the badge as shown in the ticket as such command
<hkaiser> ahh, do we really need this?
<diehlpk_work> Yes, if we cff is not correct it will not be shown on our github page
<hkaiser> diehlpk_work: I added a comment demonstrating how to add this to an rst file
<diehlpk_work> the python tool generating the cff file from the zenodo doi is so error prone, I had to fix 10 issues bu hand
<ms[m]> diehlpk_work: 👍
<ms[m]> hkaiser: did core fail to build or "only" tests/examples? We don't build everything on the windows ci
<diehlpk_work> hkaiser, ms[m] I added the badge and I think the pr is good as it is
<hkaiser> ms[m]: full fails to build
<hkaiser> ms[m]: yep: the windows CI shows the same problems: https://github.com/STEllAR-GROUP/hpx/runs/3596298644?check_suite_focus=true
<hkaiser> hmmm, not really, it has only the inconsistent dll linkage warnings, but not the duplicate symbol errors I'm seeing...
diehlpk has joined #ste||ar
diehlpk has quit [Client Quit]
nanmiao has quit [Quit: Client closed]
diehlpk has joined #ste||ar
diehlpk has quit [Ping timeout: 260 seconds]
Yorlik has quit [Read error: Connection reset by peer]
hkaiser has quit [Quit: Bye!]