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/
jaafar has joined #ste||ar
<heller> hmm, did something on rostam went awfully wrong?
<heller> all release builds seem to go horribly wrong
<K-ballo> hkaiser: making those sfinae do make the error go away.. so I guess the question is who is attempting to instantiate them
<K-ballo> was the optional PR reverted? or did it just not include replacement for uses of boost::optional?
<heller> it did not include any replacements
<github> [hpx] sithhell force-pushed fix_action_move_semantics_regression from 9c661f7 to 6ea1947: https://git.io/vbYHx
<github> hpx/fix_action_move_semantics_regression 6ea1947 Thomas Heller: Fixing local direction function execution and lambda actions perfect forwarding
<heller> for some reasons, the papi counters fail, but only in release mode and on rostam
<heller> all other plugins seem to work
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
<heller> the only difference to my local setup is that I have papi 5.5 and rostam 5.2
<heller> Ok, this problem seems to be around since a while...
mcopik has quit [Ping timeout: 255 seconds]
jaafar_ has joined #ste||ar
jaafar has quit [Ping timeout: 255 seconds]
jaafar_ has quit [Ping timeout: 255 seconds]
<hkaiser> heller: papi counters are dynamically loaded
<hkaiser> nobody links against them
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
diehlpk has joined #ste||ar
hkaiser has quit [Quit: bye]
diehlpk has quit [Ping timeout: 260 seconds]
K-ballo has quit [Quit: K-ballo]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
nanashi55 has quit [Ping timeout: 258 seconds]
nanashi55 has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
mcopik has joined #ste||ar
fane_faiz1 has joined #ste||ar
mcopik has quit [Ping timeout: 248 seconds]
mcopik has joined #ste||ar
fane_faiz1 is now known as fane_faiz
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
<fane_faiz> Hey guys i am new here.
<fane_faiz> wanted some assistance.
<github> [hpx] StellarBot pushed 1 new commit to gh-pages: https://git.io/vbOUR
<github> hpx/gh-pages cb664f3 StellarBot: Updating docs
fane_faiz has quit [Ping timeout: 248 seconds]
taeguk has joined #ste||ar
<github> [hpx] taeguk opened pull request #3040: Fix a mistake about exception handling in asynchronous execution of scan_partitioner. (master...tg_fix_scan_partitioner_async) https://git.io/vbOTk
<taeguk> Excuse me, I have a question about hpx::future::has_exception.
<taeguk> When the task is running so the result is not available from hpx::future and I call hpx::future::has_exception, it returns 'false'.
<taeguk> Is this intended behavior?
<taeguk> I assumed that hpx::future::has_exception may internally wait for the result to be available. But my assumption was incorrect.
fane_faiz has joined #ste||ar
<fane_faiz> Hello guys i wanna contribute so how should i get started?
mcopik has quit [Ping timeout: 260 seconds]
K-ballo has joined #ste||ar
mcopik has joined #ste||ar
<taeguk> fane_faiz: As I think, at first, investigate and determine what you will improve or implement.
<fane_faiz> can you please elaborate?
<github> [hpx] sithhell closed pull request #3038: Fixing local direction function execution and lambda actions perfect forwarding (master...fix_action_move_semantics_regression) https://git.io/vbYVQ
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
<zao> fane_faiz: We don't have much in the way of prepared tasks for newcomers. Skim through the SoC suggestions maybe, or look at the issue tracker for things to look at.
<zao> fane_faiz: Whatever you do, it's probably a good idea to download and build HPX, to get a feeling for how to work with it.
<zao> It all really depends on what kind of person you are and what you're capable of doing.
<zao> If you're unsure what HPX actually does, the presentations at conferences made by contributors may also be of interest, to get a feeling for what HPX actually is.
EverYoung has joined #ste||ar
fane_faiz has quit [Ping timeout: 248 seconds]
EverYoung has quit [Ping timeout: 255 seconds]
hkaiser has joined #ste||ar
<jbjnr> hkaiser: heller pull request are now being built at cscs http://cdash.cscs.ch/index.php?project=HPX - so far I have only added one machine/configuration, but my simple pything based github query tools can spawn as many builds an any machines as we like.
<jbjnr> ps. ignore the older ones with all fails, I submitted a bunch of bad ones earlier ...
<hkaiser> jbjnr: nice!
<hkaiser> jbjnr: btw, #3039 is what you need for your executor work
fane_faiz has joined #ste||ar
<zao> jbjnr: Nifty!
taeguk has quit [Quit: Page closed]
<heller> jbjnr: cool! Is there also a github status report for the PRs?
<jbjnr> hkaiser: heller currently I have a very simple python script that polls github every N seconds (60 default) and if a PR has changed, it triggers the build on a machine at cscs. It is version1 and is fire and forget, so I am not getting feedback from the builds that I can use to update the PR status, but I will add that stuff if this proves useful. I need to add several other things first.
<jbjnr> it spawns builds using ssh and a script - currently we have a sngle config per machie and all is done via cmake and python. nothing else - it is very trivial, easy to extend and maointain and I have tried it on 3 machines - daint, greina and my laptop
<hkaiser> jbjnr: nice
<jbjnr> the ssh spawns a slurm job (generated by cmake), so it will run anywhere, but daint can't send http out from compute nodes, so we need something extra for that one. I will work on that
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
<heller> jbjnr: I think that's very useful
<jbjnr> if someone needs to update a PR, then please do it now - I want to see if a new build is triggered correctly. So far I tested it using a force option - so I have not seen it happen in the wild
<jbjnr> also I need to trigger the PR's to rebuild if master changes. That's easy too
<jbjnr> I can do many build per PR, so far just one release build with my default settings, but I will add more.
<hkaiser> jbjnr: do you build the plain PR or the PR merged to master?
<jbjnr> PR merged to master
<hkaiser> nice
<zao> Do you people know what the refs/pull/ID/merge thing means?
<jbjnr> but only if the github PR status is "mergeable"
<heller> No
<zao> refs/pull/ID/head is the PR head itself, is the merge what you'd get if merged into master or something?
<heller> jbjnr: instead of polling, you might want to consider the github web hooks
<heller> zao: I think it's the code as proposed
<jbjnr> that would be nice, but then I need to open ports at CSCS to let stuff in and write an http server. the pygithub code is a few lines and so easy ....
<jbjnr> I have called it "Pycicle"
<jbjnr> Python Continuous Integration Command Line Engine
<jbjnr> :)
<heller> Yeah, polling is good enough, I guess
<jbjnr> it sleeps the rest of the ime
<heller> Nice name
<hkaiser> yah polling once every minute is fine
<zao> Ah, 'merge' seems to be a suggested merge result of merging the PR head into the master commit it's based on.
<jbjnr> I will create a PR with pycicle and see if a build is triggered
<zao> That's useful I guess.
<hkaiser> popsicle ;)
eschnett has joined #ste||ar
EverYoun_ has joined #ste||ar
EverYoun_ has quit [Remote host closed the connection]
EverYoun_ has joined #ste||ar
EverYoun_ has quit [Remote host closed the connection]
EverYoun_ has joined #ste||ar
EverYoun_ has quit [Remote host closed the connection]
EverYoun_ has joined #ste||ar
EverYoun_ has quit [Remote host closed the connection]
EverYoung has quit [Ping timeout: 255 seconds]
EverYoun_ has joined #ste||ar
EverYoun_ has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
<github> [hpx] biddisco opened pull request #3042: PyCicle - first attempt at a build toold for checking PR's (master...pycicle) https://git.io/vbOC4
jaafar_ has joined #ste||ar
<hkaiser> jbjnr: I know you will think I'm just annoying, but could you add license and copyright to the new files, please?
<jbjnr> ok
<jbjnr> build was triggered ok
<hkaiser> :D
<jbjnr> and you are correct. you are annoying :)
<hkaiser> that's what I'm here for ;)
fane_faiz has quit [Ping timeout: 248 seconds]
Smasher has joined #ste||ar
<zao> jbjnr: For running the builds and tests on the clusters, do you limit your jobs to one per node to avoid the port collisions?
<zao> Or have you altered the test running thingie to jiggle the ports around to try to counteract that?
<zao> (I run my single-node tests inside of a network-insulated singularity container)
<jbjnr> zao: no, I use slurm to submit a job to the nodes - so far, this works - but some kid of limit of <N could be enforced. The script is very very simple so far, but we can add more features. I'll send an email to the list with my plans ...
<jbjnr> on greina - it seems like 8 jobs run at a time without problem, more than that get queued - which is fine
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
mcopik has quit [Ping timeout: 248 seconds]
hkaiser has quit [Read error: Connection reset by peer]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
hkaiser has joined #ste||ar
Bibek has quit [Quit: Leaving]
Bibek has joined #ste||ar
mcopik has joined #ste||ar
<jbjnr> grrr... I don't know how to get the branch name from the PR on github
<jbjnr> I'd like to use it in the dashboard display
<jbjnr> ooh. pr.head.label - give it to me - awesome!
<zao> Ah, indeed.
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
Smasher has quit [Remote host closed the connection]
Smasher has joined #ste||ar
diehlpk has joined #ste||ar
diehlpk has quit [Ping timeout: 260 seconds]
Smasher has quit [Remote host closed the connection]
<jbjnr> zao: I misunderstood your question - yes one test at a time on a given node - but slurm will give me N nodes, so I can test N PR's at once. The port collision I didn't test, but I know from past experience that running two tests on the same node will fail, so no parallel testing
<zao> Ah, peace of mind restored.
<zao> Was worrying I'd imagined everything :)
hkaiser has quit [Read error: Connection reset by peer]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]