aserio 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/
diehlpk has joined #ste||ar
deep-book-gk_ has joined #ste||ar
deep-book-gk_ has left #ste||ar [#ste||ar]
<hkaiser> K-ballo: yt?
<K-ballo> hkaiser: kinda
<hkaiser> quick question
<hkaiser> what should we do with the atomics PR?
<hkaiser> I'm confused by th elatomic
<hkaiser> -latomic
<K-ballo> figure out the proper flags
<K-ballo> I've pinged someone but haven't heard back yet, I could just ask awilliams
<hkaiser> yah, good thinking
<hkaiser> pls keep me in the loop
hkaiser_ has joined #ste||ar
hkaiser has quit [Read error: Connection reset by peer]
hkaiser_ has quit [Client Quit]
hkaiser has joined #ste||ar
mars0000 has quit [Quit: mars0000]
mars0000 has joined #ste||ar
bikineev has joined #ste||ar
mars0000 has quit [Quit: mars0000]
vamatya has quit [Ping timeout: 276 seconds]
mars0000 has joined #ste||ar
diehlpk has quit [Ping timeout: 248 seconds]
hkaiser has quit [Quit: bye]
mars0000 has quit [Quit: mars0000]
mars0000 has joined #ste||ar
mars0000 has quit [Client Quit]
bikineev has quit [Remote host closed the connection]
vamatya has joined #ste||ar
diehlpk has joined #ste||ar
K-ballo has quit [Quit: K-ballo]
diehlpk has quit [Remote host closed the connection]
vamatya has quit [Ping timeout: 260 seconds]
diehlpk has joined #ste||ar
diehlpk has quit [Remote host closed the connection]
<wash[m]> Whats the issue number for that? K-ballo and hkaiser
patg_ has quit [Read error: Connection reset by peer]
mars0000 has joined #ste||ar
mars0000 has quit [Client Quit]
pree_ has joined #ste||ar
patg has joined #ste||ar
patg is now known as Guest11276
vamatya has joined #ste||ar
Guest11276 has quit [Quit: This computer has gone to sleep]
Guest11276 has joined #ste||ar
Guest11276 has quit [Quit: This computer has gone to sleep]
mcopik has joined #ste||ar
mcopik has quit [Ping timeout: 260 seconds]
vamatya has quit [Ping timeout: 260 seconds]
bikineev has joined #ste||ar
taeguk has joined #ste||ar
<github> [hpx] taeguk opened pull request #2788: Adapt parallel::is_heap and parallel::is_heap_until to Ranges TS. (master...tg_is_heap_range) https://git.io/v7lCt
taeguk has quit [Quit: Page closed]
<github> [hpx] StellarBot pushed 1 new commit to gh-pages: https://git.io/v7lCY
<github> hpx/gh-pages 2d8fd0a StellarBot: Updating docs
bikineev has quit [Remote host closed the connection]
pree_ has quit [Ping timeout: 240 seconds]
bikineev has joined #ste||ar
pree_ has joined #ste||ar
K-ballo has joined #ste||ar
<wash> K-ballo:
<wash> K-ballo: what's the issue number for the atomics stuff?
mcopik has joined #ste||ar
denis_blank has joined #ste||ar
<K-ballo> wash: there's no corresponding issue. I ran into it while applying changes to a branch, I put those changes on hold for now
<wash> gotcha
<K-ballo> let me look for the last CI build
<wash> what is the pr number?
<K-ballo> wash: https://circleci.com/gh/STEllAR-GROUP/hpx/7465, branch std-atomic (not std-atomic-lite)
<K-ballo> the boost lockfree replacement code ends up instantiating an atomic for a non-trivially-copyable type, std::atomic diagnoses
mcopik has quit [Ping timeout: 248 seconds]
pree_ has quit [Ping timeout: 240 seconds]
pree_ has joined #ste||ar
eschnett has joined #ste||ar
bikineev has quit [Remote host closed the connection]
hkaiser has joined #ste||ar
denis_blank has quit [Quit: denis_blank]
bikineev has joined #ste||ar
pree_ has quit [Ping timeout: 240 seconds]
bikineev has quit [Remote host closed the connection]
bikineev has joined #ste||ar
<ABresting> What's this error? https://www.irccloud.com/pastebin/AfzAQBRt/
<hkaiser> ABresting: well, file not found
<hkaiser> have you followed the documentation explaining how to build hpx?
<ABresting> yes like a 1000 times
<ABresting> and even CI didn't do this error
<ABresting> when I build it there
<hkaiser> k, then I'd ask you to give s the concrete command lines you used to build hpx, best would be a shell log
<hkaiser> ABresting: ci does it correctly
<hkaiser> so it doesn't fail
<ABresting> cmake <hpx_sourcec_tree> -DHPX_WITH_STACKOVEFLOW_DETECTION=ON
<ABresting> from the build directory
<hkaiser> what was the output
<ABresting> the one which is posted above
<hkaiser> no
<hkaiser> what did cmake output?
<hkaiser> you said you did it 1000 times, what did you do differently this time
<hkaiser> ?
bikineev has quit [Remote host closed the connection]
mcopik has joined #ste||ar
<hkaiser> ABresting: ok, now please do a make VERBOSE=1 and show me what this prints
<hkaiser> ABresting: also, please do an 'ls /home/arawat/myhpx/hpx/config'
<ABresting> hkaiser: hold on to that my laptops wifi i
<ABresting> is having troubles
<hkaiser> ABresting: frankly I doubt you have ever compiled hpx
<hkaiser> at least not the code you submitted as the PR
<ABresting> not with the last change I made, do you think hpx is that hard to compile even?
<ABresting> I am going through a tough time but you say I am listening
<hkaiser> ABresting: it is not difficult to compile
<ABresting> then dont doubt it
<hkaiser> shrug, all I know is that the PR has invalid code in it right now, and that previous versions of the PR had invalid code as well which I pointed out to you
<hkaiser> that tells me that you have never actually compiled and run the code you're trying to deliver
<hkaiser> also, despite the requirement not to add any GPL'ed code you simply copied the libsegev test code into the PR, you even left the GP: license in there...
<ABresting> I know its going to, is it poasible if CI has built it successfully and still it will suck on a cluster?
<ABresting> its inspired from it
<hkaiser> ABresting: the CI does not enable your configuration constant, does it?
<ABresting> no
<ABresting> I see your point let me fix this
<hkaiser> so why do you expect it to compile your code, then?
<zao> Which branch are we on?
<mcopik> ABresting: you can't make a copy paste of someone's code and call it 'inspiration'
<hkaiser> zao: #2783
<ABresting> its a wrapper code I cant make it any different and If i try it will be trival
<ABresting> I know coz I have tried
<hkaiser> ABresting: you even left the GPL license text in the file - we've asked you from day one to stay away from GPL
<ABresting> that I was trying to fix the config.h, I didnt know that even config file code is dangeroua
<ABresting> My bad
<ABresting> at last day I should not try and do stuff, lesson learned
<mcopik> ABresting: regarding CMakeLists, you should add yourself to the copyright list after making a meaningful contribution i.e. add at least 10-15% of new code. that's how Hartmut explained me this in the past
<ABresting> Thank you and I am deeply sorry for that I will be beyond careful the next time
<mcopik> ABresting: np :)
<zao> Note that you've spelled -DHPX_WITH_STACKOVEFLOW_DETECTION=ON wrong above.
<ABresting> ?
<ABresting> this is what happens when you are not in the constant touch of your mentor. doomsday
<hkaiser> ABresting: from what wash wrote this is not true
<mcopik> ABresting: have you compiled this code successfully on your machine? PR is for a code which can be built, has been tested and is ready to be merged with the repository. a quick glance of your PR suggests that this is not even close to being ready
<ABresting> hkaiser: I have talked to you more than I did with wash
<ABresting> but I should accept the fact that my gsoc is over
<zao> Very ugly outputs, but that's the three cases I could test.
<zao> No feature flag enabled and no sigsegv around, feature flag enabled but no sigsegv around, and feature flag enabled and libsigsegv installed.
<zao> (guess I could also test libsigsegv installed but no flag requested)
<zao> Note the <config.h> failure for the supposedly happy path.
<zao> This is of course on my bog standard debian home machine, not some weirdo cluster.
<hkaiser> thanks zao
<hkaiser> ABresting: exactly what I thought.... things don't even configure properly, mostly
<zao> The upside of using 'script' is that it sees exactly everything you type into the terminal, including all the mistakes and colour codes from scripts :)
<zao> hkaiser: At least it's harmless unless explicitly enabled with the CMake define.
<zao> It stays out of the way even when the library is on the system.
<zao> But yes, there's things to fix :)
<hkaiser> ok
bikineev has joined #ste||ar
<github> [hpx] ABresting closed pull request #2783: StackOverflow integration with libsigsegv (master...master) https://git.io/v7Oid
diehlpk has joined #ste||ar
hkaiser_ has joined #ste||ar
hkaiser has quit [Read error: Connection reset by peer]
pree_ has joined #ste||ar
pree_ has quit [Ping timeout: 276 seconds]
<wash[m]> ABresting: to clarify - and for the benefit of other's - the issue isn't /using/ libsigsegv, that's fine. The trouble is that the tests that you used as "inspiration" were GPL licensed as well; so that means your code is GPLd. If you wrote code from scratch that used the library (just using the library docs), it would have been fine.
<ABresting> I will restart from scratch this time. seriously didnt know that test code will also get in the gpl way but good thing is I know what to do.
pree_ has joined #ste||ar
diehlpk has quit [Ping timeout: 260 seconds]
<heller> do we know where the red/orange in buidbot is coming from?
<K-ballo> "Move the unwrapped tests from lcos to util" apparently
<heller> ok
<K-ballo> some bad 11-constexpr that should have been 14
patg has joined #ste||ar
<heller> ok, so should be easy enough to fix, i guess
<K-ballo> trivial
diehlpk has joined #ste||ar
pree_ has quit [Quit: AaBbCc]
pree_ has joined #ste||ar
<K-ballo> you guys knew there's an #hpx channel on the cpp slack?
<pree_> On slack really ??
<patg> K-ballo: I asked for one but didn't know they actually made one
<patg> and for some reason I can't join it from my work computer
<patg> guess I have to be on location not on a work laptop at home I'll try tomorrow
<K-ballo> patg: I suspect you are referring to a different cpp slack, I know you guys have an internal one
<heller> K-ballo: No, I didn't know
<heller> Let me check
<K-ballo> patg: I meant this one https://cpplang.now.sh/
<heller> There's no action in it
<K-ballo> cpplang.slack.com
<K-ballo> nothing in it, but note there was one "user" in there already and none of us
patg has quit [Quit: See you later]
<heller> Yeah
pree_ has quit [Quit: AaBbCc]
<wash[m]> K-ballo: I believe I made a stellar channel on slack too
<github> [hpx] hkaiser force-pushed resource_partitioner from 7467960 to 9b8e3e1: https://git.io/v7lfK
<github> hpx/resource_partitioner 9b8e3e1 Hartmut Kaiser: Adding pool specific performance counters...
<K-ballo> wash[m]: oh yeah, that too... just by /join I suppose they materialize?
<wash[m]> Yah, I think I camped stellar so we would have one we could control. Dunno who made hpx
<wash[m]> First one there is admin on slack
<K-ballo> :)
<zao> Clearly those dastardly HPX5 people.
<zao> Or is it HPX3? :P
deep-book-gk_ has joined #ste||ar
deep-book-gk_ has quit [Excess Flood]
<K-ballo> hkaiser_: wrt -latomic, long story short libc++/libstdc++ implement std::atomic via _Atomic, which in turn may emit calls to libatomic, in particular for those operations that aren't lockfree on a given platform
<K-ballo> here's a bit more info https://gcc.gnu.org/wiki/Atomic/GCCMM/LIbrary
diehlpk has quit [Ping timeout: 240 seconds]
<hkaiser_> K-ballo: ok
<hkaiser_> so we need to link with -latomic in the feature test?
<K-ballo> yes, as far as I understand, if not msvc then we should add -latomic since it _may_ be needed
<K-ballo> that reminds me, I assume you haven't noticed this line before https://github.com/STEllAR-GROUP/hpx/blob/master/cmake/HPX_AddConfigTest.cmake#L474
<hkaiser_> right I have not
<hkaiser_> K-ballo: do we need to add -latomic to the actula build as well, then?
<hkaiser_> if not msvc, that is
<K-ballo> yes
<hkaiser_> nod
<hkaiser_> ok, let's do that, then
<K-ballo> for weird platforms, or atomic<big-UDT>, etc
<K-ballo> we shouldn't have atomic<big-UDT>s, because that uses kernel mutexes.. but...
<K-ballo> who knows
Matombo has joined #ste||ar
bikineev has quit [Remote host closed the connection]
<zao> Makes some grand claims of being stupidly fast for stuff.
bikineev has joined #ste||ar
taeguk[m] has quit [Ping timeout: 246 seconds]
thundergroudon[m has quit [Ping timeout: 246 seconds]
<hkaiser_> zao: I trust a memory manager only starting V3 or so ;)
<K-ballo> ok, so where would -latomic go?
<hkaiser_> K-ballo: proximity of -pthread?
<K-ballo> .....uh, ehm
Matombo has quit [Remote host closed the connection]
Matombo has joined #ste||ar
<K-ballo> I'm pretty sure we are just implicitly inheriting that from boost... if at all...
Matombo has quit [Remote host closed the connection]
<K-ballo> where did the code that was adding -stdlib=libc++ go?
mcopik has quit [Ping timeout: 260 seconds]
thundergroudon[m has joined #ste||ar
taeguk[m] has joined #ste||ar
Matombo has joined #ste||ar
<K-ballo> how do I blame-forward?
<K-ballo> wow, apparently it's been gone since at least mid-2014 :|
<K-ballo> anyways, I suppose `hpx_libraries(atomic)` would be the expected thing to do?
diehlpk has joined #ste||ar
Matombo has quit [Remote host closed the connection]
Matombo has joined #ste||ar
<github> [hpx] K-ballo force-pushed std-atomic-lite from 32af53d to baf4deb: https://git.io/v7tLl
<github> hpx/std-atomic-lite 3f6c668 Agustin K-ballo Berge: Replace boost::atomic with std::atomic (where possible)
<github> hpx/std-atomic-lite baf4deb Agustin K-ballo Berge: Add deprecated checks for Boost.Atomic
<K-ballo> this looks like a more appropriate place https://github.com/STEllAR-GROUP/hpx/blob/master/CMakeLists.txt#L1138
<K-ballo> although the two uses of -pthread might be distinct
<hkaiser_> yah
<hkaiser_> is there a way to check whether libatomic exists?
<K-ballo> find_library? try_compile?