hkaiser 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/ | GSoC: https://github.com/STEllAR-GROUP/hpx/wiki/Google-Summer-of-Code-%28GSoC%29-2020
kale[m] has quit [Ping timeout: 260 seconds]
kale[m] has joined #ste||ar
Yorlik has quit [Ping timeout: 240 seconds]
hkaiser has quit [Quit: bye]
kale[m] has quit [Ping timeout: 240 seconds]
kale[m] has joined #ste||ar
bita_ has quit [Ping timeout: 260 seconds]
nikunj97 has joined #ste||ar
Nikunj__ has joined #ste||ar
nikunj97 has quit [Ping timeout: 260 seconds]
Nikunj__ has quit [Read error: Connection reset by peer]
kale[m] has quit [Read error: Connection reset by peer]
kale[m] has joined #ste||ar
nikunj97 has joined #ste||ar
<nikunj97> ms[m], yt?
Yorlik has joined #ste||ar
nikunj97 has quit [Read error: Connection reset by peer]
<gonidelis[m]> ms[m]: K-ballo how did you solve the `include force_linking.hpp` problem on `modules.cpp` ?
<gonidelis[m]> I 've just rebased to master
nikunj97 has joined #ste||ar
Nikunj__ has joined #ste||ar
nikunj97 has quit [Ping timeout: 240 seconds]
Nikunj__ has quit [Read error: Connection reset by peer]
<gonidelis[m]> I saw that you said "with a clean cache" it works... How could I clean the cache?
<K-ballo> gonidelis[m]: wipe your entire build folder, start from scratch
<gonidelis[m]> yeah... imagined so. Just did it. Thanks a lot.
hkaiser has joined #ste||ar
<gonidelis[m]> @vscode_users any idea how could I make vscode indicate-highlight the angle brackets pair in a cpp/hpp file?
<gonidelis[m]> hkaiser: K-ballo ^^ tried swapping the two lines but didn't work. Any idea where the problem might be?
<K-ballo> I was not expecting the swap to work, just to give us more insight on the error... but that's not the runner that had that error
<K-ballo> that one is a genuine failure, and looks to be from a different branch?
kale[m] has quit [Ping timeout: 265 seconds]
kale[m] has joined #ste||ar
<gonidelis[m]> From a different branch?
<gonidelis[m]> Yeah... actually this particular error came after changing `!disable_sized_sentinel_fro<Sent, Iter>` to `!disable_sized_sentinel_fro<typename remove_cv<Sent>::type, typename remove_cv<Iter>::type>`
<K-ballo> for good reason
<hkaiser> gonidelis[m]: it's std::remove_cv<>
<gonidelis[m]> ahhh that was stupid of me :@ hkaiser thanks
<gonidelis[m]> https://github.com/STEllAR-GROUP/hpx/pull/4768/files I can see some changes that are listed here but are part of a previous commits which have been merged by now... Is it a ok if i delete those files from the github interface?
<hkaiser> gonidelis[m]: if you rebase your branch on master those should go away
<gonidelis[m]> So if I rebase shouldn't I git rm those files before commiting?
<gonidelis[m]> hkaiser: ^^
<K-ballo> after a rebase those files will no longer have changes
<K-ballo> those changes you see now would be associated with existing commits
<gonidelis[m]> I suspect that this happened because I tried to avoid `git push --force`
<gonidelis[m]> And git showed me "error: failed to push some refs to 'https://github.com/gonidelis/hpx.git' "
<gonidelis[m]> And then I `git pull`ed before pushing
<K-ballo> as long this is a PR being worked on force pushing is fine, encouraged even, although consider –force-with-lease instead
nikunj97 has joined #ste||ar
<gonidelis[m]> Ok I will look up what force-with-lease does and how it works. Thanks ;)
Nikunj__ has joined #ste||ar
<gonidelis[m]> hkaiser: Every changed-file now must be completely correct. I am expecting for the build tests to succeed :) There is only this unit test for `is_sized_sentinel_for` missing right now. I will add it asap and I think this PR should be complete then :D :D :D (finally!)
nikunj97 has quit [Ping timeout: 260 seconds]
Nikunj__ is now known as nikunj97
<nikunj97> hkaiser, how is the libfabric support within hpx?
<hkaiser> nikunj97: flaky at best
<nikunj97> hkaiser, so use MPI and make MPI use Infiniband
<nikunj97> ?
<hkaiser> yes
<nikunj97> alright!
weilewei has joined #ste||ar
<weilewei> wow, Hazard pointers related unit and stress tests all passed (all container types) with HPX support now
<weilewei> I used std::Vector<size_t> to store different types of thread data in hpx. Will using map<int, size_t> makes more sense?
<gonidelis[m]> K-ballo: Shouldn't an `input iterator` work as a sized sentinel for an `input iterator` ?
<K-ballo> only if random access
<hkaiser> weilewei: depends on what you want to store?
<gonidelis[m]> right... missed it once again. Thanks! So random-access iterator pair should pass is_sized_sentinel ... are there any non-custom pairs that should pass?
<gonidelis[m]> K-ballo:
<hkaiser> how many size_t's do you need?
<hkaiser> gonidelis[m]: vector<T>::iterator should pass
<hkaiser> vector<T>::const_iterator should pass too, so should std::string::iterator
<gonidelis[m]> great! Thanks!
<gonidelis[m]> Isn't that the type of a random access iterator indicated by std::random_access_iterator_tag ??
<K-ballo> I can't quite parse that question, but I'd guess yes?
<gonidelis[m]> Rephrasing: Does std::random_access_iterator_tag indicate a type?
<K-ballo> a "type" in what meaning of the word?
<gonidelis[m]> a typename
<K-ballo> I have no idea what your underlying question i
<K-ballo> is
<K-ballo> random_access_iterator_tag is a type itself, a struct type.. that doesn't seem relevant here
<gonidelis[m]> ok sory... I understand your point. I wanted to ask how could I represent a random access iterator like `std::random_access_iterator randIter`. But I don't think there is such thing as `std::random_access_iterator_type` or sth....
<K-ballo> no, there isn't one
<gonidelis[m]> great...
<K-ballo> there's no one single random access iterator type, random access is a concept, it's a family of types
<gonidelis[m]> So ther is only an implicit way of creating a random access iterator
<gonidelis[m]> hmmm "a family of types" ... so the members of this family should be `difference_type, value_type, pointer, reference ,iterator_category` . Right??
<K-ballo> no, members of that family are std::string::iterator, vector<T>::iterator, vector<T>::const_iterator, and so on (it's unbounded)
<weilewei> hkaiser so far just two size_t plus one comes with default hpx thread_data size_t
<weilewei> so in total three
<gonidelis[m]> damn...
<weilewei> so far I just want to store size_t, not a copy of libcds thread data structure
<weilewei> *store pointer to size_t
<gonidelis[m]> hkaiser: K-ballo just added an is_sized_sentinel_for test. Please leave your feedback on the PR conversation (won't be available on IRC for the next couple hours) on what other pairs should I check and if there is anything that you think should be changed :) Thanks!
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
<hkaiser> weilewei: please leave the original thread_data (size_t) alone
<hkaiser> also, I'd suggest not to use a vector<> as it requires allocation which we want to avoid, if possible
nikunj97 has quit [Read error: Connection reset by peer]
bita_ has joined #ste||ar
nikunj97 has joined #ste||ar
nikunj has quit [Ping timeout: 246 seconds]
Nikunj__ has joined #ste||ar
nikunj97 has quit [Ping timeout: 244 seconds]
<hkaiser> kale[m]: correct
<hkaiser> so the tests are never run ;-)
<Nikunj__> parsa, were these the same tests you were talking about?
<Nikunj__> I remember you mentioning failing tests in the last call with kale[m]
Nikunj__ is now known as nikunj97
<parsa> Nikunj__: nothing works really even the simplest things… it's not any specific feature or test that's failing
<nikunj97> so it's just import phylanx that works?
<nikunj97> kale[m] ^^
<parsa> import phylanx loads phylanx's python front-end… that works, yes. also the phylanx python binding loads successfully
<parsa> but the libraries that phylanx itself tries to load don't load
<nikunj97> I see
<nikunj97> kale[m], any luck correcting the rpath?
<kale[m]> I've updated the rpath of the shared libraries. When I run the test it throws an error FATAL: exception bot rethrown
Nikunj__ has joined #ste||ar
nikunj97 has quit [Ping timeout: 260 seconds]
<kale[m]> find . -name "*.so*" | xargs -I{} ldd {} | grep "not found" in the installation location shows no result ; so all rpath are setup correctly. All the libs that phylanx builds in /libs are included in the package as well. I will compare the installation using wheel vs build to find out the underlying problem tomorrow.
Nikunj__ has quit [Read error: Connection reset by peer]
diehlpk_work_ has joined #ste||ar
diehlpk_work has quit [Ping timeout: 260 seconds]
<K-ballo> silly msvc doesn't know how to do typeid in a constant expression... I reported that bug ages ago
<K-ballo> I was thinking of bringing back the old type_info based function implementation, for dll support, and to get rid of the extra member
hkaiser has quit [Ping timeout: 240 seconds]
hkaiser has joined #ste||ar
<hkaiser> gonidelis[m]: the circleci test failure is unrelated, I'm sure - even if I have not seen this before
kale[m] has quit [Ping timeout: 260 seconds]
kale[m] has joined #ste||ar
hkaiser has quit [Ping timeout: 258 seconds]
hkaiser has joined #ste||ar