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/
hkaiser has joined #ste||ar
hkaiser has quit [Quit: bye]
akheir has quit [Remote host closed the connection]
diehlpk_work has joined #ste||ar
diehlpk_work_ has quit [Ping timeout: 264 seconds]
hkaiser has joined #ste||ar
gnikunj[m] has quit [Ping timeout: 240 seconds]
k-ballo[m] has quit [Ping timeout: 240 seconds]
gonidelis[m] has quit [Ping timeout: 240 seconds]
parsa[m] has quit [Ping timeout: 240 seconds]
jpinto[m] has quit [Ping timeout: 240 seconds]
tiagofg[m] has quit [Ping timeout: 246 seconds]
pedro_barbosa[m] has quit [Ping timeout: 240 seconds]
mdiers[m] has quit [Ping timeout: 260 seconds]
klaus[m] has quit [Ping timeout: 260 seconds]
teonnik has quit [Ping timeout: 268 seconds]
rori has quit [Ping timeout: 240 seconds]
ms[m] has quit [Ping timeout: 246 seconds]
ms[m] has joined #ste||ar
gonidelis[m] has joined #ste||ar
rori has joined #ste||ar
klaus[m] has joined #ste||ar
pedro_barbosa[m] has joined #ste||ar
heller1 has joined #ste||ar
gnikunj[m] has joined #ste||ar
teonnik has joined #ste||ar
tiagofg[m] has joined #ste||ar
mdiers[m] has joined #ste||ar
jpinto[m] has joined #ste||ar
k-ballo[m] has joined #ste||ar
parsa[m] has joined #ste||ar
<ms[m]> hkaiser: up early again? ;) the sanitizer failure is pretty reproducible, but I don't know what's going on
<ms[m]> I have a stacktrace which might tell you something, there's no obvious (to me) connection to the changes in the PR
<hkaiser> interesting
<hkaiser> it's that segfault, right?
<ms[m]> actually, reinitializable_static...
<hkaiser> uggh
<hkaiser> I can't reproduce this locally :/
<hkaiser> so I can't actually bisect it
<ms[m]> I'll see what reverting that reinitializable_static commit does
<ms[m]> I'm running set_thread_state_test btw
<gonidelis[m]> is there any way to compile all tests that have to do with `find` like `make tests.unit.modules.algorithms.find*` or `%` idk, sth?
<hkaiser> gonidelis[m]: the ctest command line supports regexes, I believe
<hkaiser> ms[m]: ok, I'll try that test
<gnikunj[m]> hkaiser wow, you're up early again!
<gonidelis[m]> but the `make` one doesn't
<hkaiser> ms[m]: working on #5113 now, btw
<gonidelis[m]> gnikunj[m]: i am not even surprised any more
<ms[m]> gonidelis[m]: nope... but ninja/make $(ninja/make help | grep "whatever you want") is the closest substitute
<hkaiser> gonidelis[m]: you should be able to script it, make help lists all targets, grep them and generate a list
<ms[m]> and some cut-ting thrown in
<gonidelis[m]> ms[m]: interesting. I will work it out then. (you knew it was ninja ;p )
<ms[m]> hkaiser: same with that commit reverted... I'll try bisecting but iirc not all commits compiled last time I tried
<hkaiser> :/
<mdiers[m]> hkaiser: Good morning. Did you have a look at the last callstack of #5117 from the segfault in https://github.com/STEllAR-GROUP/hpx/blob/39e39b63f93a48dc8938b68336eafbacb706b8d4/libs/core/execution_base/src/agent_ref.cpp#L53 yesterday?
<hkaiser> mdiers[m]: yah, it's the same as before
<hkaiser> it just has different functions inlined
<hkaiser> mdiers[m]: does the problem happen in debug mode as well?
<mdiers[m]> hkaiser: Yes, but due to the runtime, it is extremely time consuming to get it to crash.
<hkaiser> I understand
<hkaiser> it might give us more information if you could invest the time to get a stack backtrace from the crash in debug
<hkaiser> but I don't want to impose
<mdiers[m]> hkaiser: Is soon weekend, in the time it can then run. Which optimization parameters are desired: HPX with CMAKE_BUILD_TYPE=Debug and the application with -g -o0?
Vir has quit [Quit: ZNC 1.7.5+deb4 - https://znc.in]
Vir has joined #ste||ar
<hkaiser> mdiers[m]: nod, that should do it
<hkaiser> ms[m]: is that triggering the problem?
<ms[m]> (off for lunch now though)
<ms[m]> yes
<hkaiser> ok, I'll think about it
<ms[m]> makes sense since it's touching reinitializable_static, but I haven't looked at what the problem could be yet
<hkaiser> actually, topology doesn't need to be reinitialized
<hkaiser> I remember applying this change, but I don't remember why I did it :/
<gonidelis[m]> what's the target for segmented algos?
<gonidelis[m]> `tests.unit.modules.segmented_algorithms.(sth)` does not seem to work for me
<hkaiser> gonidelis[m]: the distributed tests have their own names
<hkaiser> test.unit.modules.distributed.tcp.segmented_algorithms.* or somesuch
<gonidelis[m]> like `tests.unit.partitioned_vector_****` ?
<gonidelis[m]> ohhh....
<gonidelis[m]> ok let me check it
<gonidelis[m]> thanks
<gnikunj[m]> hkaiser: I wanted to look into the kokkos resiliency project. Anything suggestions from your end or should I just install kokkos and play with it?
<hkaiser> ahh it's test.unit.modules.segmented_algorithms.distributed.tcp.*
<gonidelis[m]> hkaiser: #5119 seems to complain on the distributed part ;/
<hkaiser> gonidelis[m]: grep for detail::fin_* and change the headers and namespaces
<gnikunj[m]> this time I'll do everything on Visual Studio as well. Should finally give me the opportunity to learn it.
<hkaiser> gnikunj[m]: yah, start with kokkos and the hpx_kokkos repo
<gnikunj[m]> alright. Sounds good!
<gonidelis[m]> hkaiser: Ok that's what i remembered. I had used some cmake flags that cut them out though that's why i couldn't find it
<hkaiser> gonidelis[m]: I don't think you can run the distributed tests with make
<gonidelis[m]> ~.^ seriously?
<hkaiser> you need to use ctest for those
<hkaiser> but you can certainly build them with make/ninja
<gonidelis[m]> hkaiser: yeah sorry. I meant building then. that's what i check anyways
<hkaiser> and for this, make help will give you the target name
<gonidelis[m]> if it builds, it probably passes
<hkaiser> most likely
<gonidelis[m]> ok, i will be roaming in the realm of segmented for some time now. I hope I get back safe
<hkaiser> it's just a namespace change, come on
<gonidelis[m]> haha, i am exaggerating don't worry
<ms[m]> hkaiser: yep, I thought it might not need reinitialization
<hkaiser> ms[m]: the change to reinitialized_static itself looks ok to me
<mdiers[m]> hkaiser: A little problem with the debug build. Undefined reference to hpx_exported_plugins_list_hpx_factory and hpx_exported_plugins_list_hpx_registry. Which lib is missing?
<ms[m]> and I think there was a leak somewhere there which probably triggered you to make that change
<ms[m]> hmm
<hkaiser> mdiers[m]: it tries to link with the release version, some mixup happened
<hkaiser> ms[m]: let's revert this commit and see what asan says
<hkaiser> I'll have a look then
<ms[m]> 👍️
<ms[m]> hkaiser, K-ballo I just merged the boost and clang minimum version bump pr, you can rebase #5031 and #5092 on master now
<hkaiser> thanks
<ms[m]> K-ballo: there should now be at least one ci configuration for each boost version between 1.66 and 1.75
<hkaiser> I'll do that for both
<hkaiser> done
<ms[m]> hkaiser: you free to join the call?
<hkaiser> ms[m]: uhh yah
<hkaiser> are you still on the call?
akheir has joined #ste||ar
<gnikunj[m]> ms: hkaiser I would like to join the HPX discussions meeting starting next time. Is there something special I need to do?
<ms[m]> gnikunj: not really
<ms[m]> are you on the hpx-devel mailing list?
<gnikunj[m]> I think so
<gnikunj[m]> just checked, I am on the list
<ms[m]> katie sends reminders usually a day before the meetings, that contains details to join
<gnikunj[m]> ok
<ms[m]> hkaiser: sorry, I was way off on the actions pr, it's just that networking is off on the configuration that's failing...
<hkaiser> uhh, did I forget to disable things again?
shahrzad has joined #ste||ar
<mdiers[m]> <hkaiser "mdiers: it tries to link with th"> Right, another version of qtcreator was installed in the container, and the debug definitions are missing. 😟
heller1 has quit [Quit: Idle for 30+ days]
<mdiers[m]> > <@freenode_hkaiser:matrix.org> mdiers: it tries to link with the release version, some mixup happened
<mdiers[m]> * Right, another version of qtcreator was installed in the container, and the debug definitions are missing. 😟
<mdiers[m]> The test is running, performance is 1/15.
<hkaiser> mdiers[m]: thanks!
<gonidelis[m]> are these fails stemming from the pr?
powderluv has joined #ste||ar
<gonidelis[m]> Plus is there anyone who uses vs code here that knows how to clang-format every cpp file on save?
<parsa> gonidelis[m]: i use vscode and clang-format but i do formats manually and usually for blocks of code
<gonidelis[m]> parsa: ahh... i haven't find anyone on internet doing it autamtically actually
<gonidelis[m]> parsa: blocks of code ?? wow... why and how ? :p
<parsa> wouldn't the `editor.formatOnSave` setting solve your problem?
<parsa> how? ctrl-k,ctrl-f
<gonidelis[m]> ctrl k invokes clangformat9 ?
<gonidelis[m]> i don't know if the default formater on vscode is the one we use
<parsa> no it invokes clang-formatt
<parsa> if you want a different executable you must configure it
<gonidelis[m]> hmm
<parsa> or vscode wouldn't know what to use
<gonidelis[m]> ok i will try it... how do i configure it exactly?
<gonidelis[m]> wow!!
<gonidelis[m]> what's that?
<parsa> found it. search for format in settings. C_Cpp: Clang_format_path
<gonidelis[m]> i am in that live session you opened
<parsa> you'd need the `ms-vscode.cpptools` plugin
<gonidelis[m]> yeah already have that
<parsa> sorry, i expected that it would always show you the window i am in. it only shows the editor apparently
<gonidelis[m]> haha no prob
<gonidelis[m]> i think your suggested setting works
<gnikunj[m]> gonidelis[m]: install the ms-vscode.cpptool plugin and the clang-format plugin (the one with the dragon). Then add the "editor.formatOnSave": true to your settings.json
<gnikunj[m]> if it still doesn't work that simply means that you haven't set the default formatter. Use ctrl+shift+f (i if on Linux) and vs code will prompt you to select the formatter
<gnikunj[m]> select clang-format there and that should do the trick
<gnikunj[m]> I haven't faced any other issues using clang-format otherwise
<gonidelis[m]> ctrl shift f
<gonidelis[m]> opens the search bar (??)
<gnikunj[m]> use ctrl shift i
<parsa> ctrl-k and then ctrl-f
<gnikunj[m]> that's for Visual Studio iirc
<gnikunj[m]> are you in vs or vscode?
<gnikunj[m]> gonidelis[m]: ^^
<parsa> same keybindings in both for me
<gonidelis[m]> vs code
<gnikunj[m]> why is it ctrl shift i for me then :/
<gonidelis[m]> gnikunj[m]: same for me
<gonidelis[m]> it worked ;)
<gnikunj[m]> nice!
<parsa> gnikunj[m]: you can choose between several options. i guess i always choose visual studio's keybindings
<gonidelis[m]> what's more in vs studio?
<gonidelis[m]> AND WHY TH MS DECIDED TO GIVE THEM TWO EXTREMELY SIMILAR NAMES?
<gonidelis[m]> (ok i am fine)
<gnikunj[m]> parsa: aah that makes sense. Do you know a good VS tutorial that teaches how to use it effectively? I'm significantly slowed down while working on anything that's more than a standalone program.
<parsa> effectively? no idea. i usually discover things while waiting on the c++ compiler :)))
<parsa> no idea about keybinding choices. also ctrl-k ctrl-f isn't just on my machine according to the docs: https://code.visualstudio.com/shortcuts/keyboard-shortcuts-windows.pdf
nanmiao11 has joined #ste||ar
nanmiao11 has quit [Client Quit]
bita has joined #ste||ar
bita has quit [Ping timeout: 260 seconds]