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/
eschnett has joined #ste||ar
jgolinowski has quit [Remote host closed the connection]
hkaiser has quit [Quit: bye]
K-ballo has quit [Quit: K-ballo]
eschnett has quit [Quit: eschnett]
david_pfander has joined #ste||ar
ste||ar-github has joined #ste||ar
<ste||ar-github> [hpx] sithhell closed pull request #3435: Fixing 1205 (master...fixing_1205) https://github.com/STEllAR-GROUP/hpx/pull/3435
ste||ar-github has left #ste||ar [#ste||ar]
ste||ar-github has joined #ste||ar
<ste||ar-github> [hpx] sithhell deleted fixing_1205 at 0d64e9a: https://github.com/STEllAR-GROUP/hpx/commit/0d64e9a
ste||ar-github has left #ste||ar [#ste||ar]
<heller> simbergm: the tss stuff is fixed. Ignore while lock is next
<simbergm> heller: you're awesome
jgolinowski has joined #ste||ar
<heller> simbergm: we should plan for a new release, I think
<heller> Traditionally for around supercomputing
simbergm has quit [Remote host closed the connection]
K-ballo has joined #ste||ar
brjsp has joined #ste||ar
<brjsp> Hello
<brjsp> I am writing with regards to this issue https://github.com/STEllAR-GROUP/hpx/issues/3175
<brjsp> The operator<< from ostream aren't virtual
<brjsp> AFAIK about the only way to customize them
<brjsp> Is to do something like subclassing streambuf, giving the customized one to ostream
<brjsp> And customizing the xsputn and overflow functions in streambuf
<brjsp> The problem concerns only the “lazy” operators, where you put a raw object to be printed, right?
<zao> brjsp: Hi there! I don't know anything about this, but figured I'd greet you at least :)
<zao> Seems like Hartmut isn't around at the moment, but someone else might have a clue.
<zao> K-ballo: Did you do anything around the op<< stuff referenced here, or know who might?
<K-ballo> it rings a vague bell
<K-ballo> no, I didn't, this is unrelated.. the virtuals mentioned in there might be the streambuf's ones
<brjsp> what is unrelated to what?
<K-ballo> my stuff to issue 3175
hkaiser has joined #ste||ar
hkaiser has quit [Client Quit]
jbjnr has quit [Quit: Leaving]
jbjnr has joined #ste||ar
hkaiser has joined #ste||ar
<jgolinowski> jbjnr, yt?
<brjsp> Hello Hartmut
<brjsp> do you know anything about this operator<< stuff?
<hkaiser> hey brjsp
<brjsp> [13:02] <brjsp> I am writing with regards to this issue https://github.com/STEllAR-GROUP/hpx/issues/3175
simbergm has joined #ste||ar
<hkaiser> ahh yes
<brjsp> should i copy the rest or do you see chat history?
<hkaiser> I can see the history, sec
<hkaiser> yes, your assessment is correct
<hkaiser> the hpx::ostream needs a major overhaul
<brjsp> The lazy operators call to streaming_operator_lazy
<brjsp> Which calls to
<brjsp> the std::ostream operators
<brjsp> Which calls to whatever streambuf we pass
<brjsp> But the problem is
<brjsp> The non-lazy operators
<brjsp> ALSO use std::ostream::operator<<
<hkaiser> yah, it's royally screwed up ;-)
<brjsp> Which means it is not possible to fix only the lazy operator without rewriting everything else
<hkaiser> nod
<hkaiser> the lazy operator is not really necessary as it's doing almost the same as the other one
<hkaiser> both are 'lazy' in some sense
eschnett has joined #ste||ar
aserio has joined #ste||ar
hkaiser has quit [Quit: bye]
brjsp has quit [Quit: Page closed]
david_pfander has quit [Quit: david_pfander]
<K-ballo> after some unexplained issues with vcpkg, I finally got to build phylanx
<K-ballo> apparently not successfully though, get an error about missing functions in symbol tables
<K-ballo> are there subtleties to working directories, plugins, etc?
<aserio> K-ballo: yes
<aserio> K-ballo: you have to build the plugins explicitly in VS
<K-ballo> I build them, they ended up in phylanx/ relative to my test binary, that ok?
jaafar has quit [Ping timeout: 252 seconds]
<aserio> K-ballo: The last time I build them it just worked out of the box..... so yes?
<aserio> :p
<aserio> I want to say that they are automatically built in the right place but there was some CMake issue with Windows which required to to explicitly build them
hkaiser has joined #ste||ar
aserio has quit [Ping timeout: 250 seconds]
aserio has joined #ste||ar
<diehlpk_work> Some one transformed the university to Super Mario Land. Was amazing
aserio has quit [Ping timeout: 252 seconds]
aserio has joined #ste||ar
jaafar has joined #ste||ar
aserio has quit [Ping timeout: 260 seconds]
<K-ballo> it's crashing in some weird ways, I'll have to build in debug mode to see what's going on
aserio has joined #ste||ar
aserio has quit [Ping timeout: 250 seconds]
jgolinowski has quit [Quit: Leaving]
aserio has joined #ste||ar
aserio has quit [Ping timeout: 240 seconds]
hkaiser has quit [Quit: bye]
bibek has quit [Quit: Konversation terminated!]
brjsp has joined #ste||ar
<brjsp> I am adjusting the find_* algorithms for Ranges TS
<brjsp> And just found a very stupid bug when find_end sometime used == istead of the user given comparison operator
bibek has joined #ste||ar
<K-ballo> missed by the tests?
<brjsp> yes
<brjsp> The tests apparently never bothered to to custom equalities
<K-ballo> :/
<brjsp> I will be doing a pull request today or tomorrow
<K-ballo> with new tests?
ste||ar-github has joined #ste||ar
<ste||ar-github> [hpx] brjsp opened pull request #3436: Add projection function to find_* (and fix very bad bug) (master...rangests) https://github.com/STEllAR-GROUP/hpx/pull/3436
ste||ar-github has left #ste||ar [#ste||ar]
<brjsp> I copied the tests over with a very dumb projection function “x -> x%65536”
<brjsp> With fixed find_end
<diehlpk_work> One question when I want to use two nodes on rostam, I just say
<diehlpk_work> 5 #SBATCH -p reno,bahram
<diehlpk_work> 6 #SBATCH -N 2
<diehlpk_work> or how do I say that I want to use reno and bahram for my job
eschnett has quit [Quit: eschnett]
<brjsp> Does anyone know about HPX_CONCEPT_REQUIRES
<brjsp> Particularly, can any code break if any of the “concept requires” are omitted, or are they just to make compiler errors less cryptic?
<brjsp> Do i really need to make up a trait expressing “is indirectly equality comparable with blah blah” for the range function to work?
<K-ballo> you.. do
<K-ballo> would you link me to the source in github?
<K-ballo> there should already be enough traits and related facilities to support your case
<brjsp> The non-range versions do not have any traits
<brjsp> You have those things like
<brjsp> line 163 — 170
<K-ballo> I'm not sure I follow you
<K-ballo> are you discussing the range versions or the non-range versions?
<brjsp> range versions
<K-ballo> there is no range find
<brjsp> Yup, am writing it
<K-ballo> oh ok, so you can't link to it yet
<K-ballo> yes, you need to do the traits
<K-ballo> and you should have enough already available
<K-ballo> so you wouldn't need to actually make up new traits
<K-ballo> unless I'm misinterpreting you?
<brjsp> How do i know which traits do i need
<K-ballo> how does your interface/spec look?
<brjsp> ignore the sentinels
<brjsp> I would need something like “is indirect comparable”
<K-ballo> there is/should be an utility "equal<>" somewhere, IIRC find is implemented as find_if(equal<>)
<K-ballo> so what you need is is_indirect_callable<... equal<>>
<K-ballo> this one: "IndirectCallableRelation<equal_to<>, projected<I, Proj>, const T*>"?
<brjsp> yes
<brjsp> I reimplemented the sequential find as
<brjsp> auto&& equality = std::bind(
<brjsp> std::equal_to<void>(), std::placeholders::_1, val);
<brjsp> return std::find_if(first, last,
<brjsp> util::invoke_projected<decltype(equality), Proj>(
<brjsp> std::move(equality),std::forward<Proj>(proj)));
<K-ballo> we can't use `std::equal_to<void>` because that's C++14, but there is a C++ implementation in the codebase already
<brjsp> Ouch
<K-ballo> just check the implementation of other algorithms, it's just a matter of replacing `std::equal_to<void>` with `util::equal_to` or something like that
<brjsp> Thanks
<brjsp> Ouch, detail::compare_to requires the type to be copyable
<brjsp> And find doesnt
<brjsp> Or is it something we dont care about because parallel and all
<brjsp> As in, can i copy val int copare_to
<brjsp> into*
<K-ballo> no, it doesn't
<brjsp> I can't do this
<brjsp> detail::compare_to<T const &>
<K-ballo> it a) requires copy or move, b) works with references too
<K-ballo> no?
<brjsp> Because then the constructors get duplicated
<K-ballo> that's odd, we have tie-breaker rules for that case in the standard
<K-ballo> T const& const& would win over T const& &&
<K-ballo> note your std::bind based predicate requires copies too
<brjsp> How would you do this function
<brjsp> detail::compare_to<T const &> does not compile under MSVC, haven't checked other compilers
<brjsp> Ugh
<brjsp> This may take a long time
<K-ballo> you could look at parallel find_if for inspiration
<brjsp> I can do a pull request with just the bug fix for find_end
<brjsp> find_if is easy
<brjsp> Or i could rangify everything except find
<K-ballo> my bad, I meant parallel find
<K-ballo> rangify-ed parallel find will end up calling parallel find underneat anyhow, won't it?
<brjsp> Nnnnno
<brjsp> You see, the whole problem is how to put the value into find_if
<K-ballo> no? mmh
<brjsp> Should we just use a freaking capturing lambda? That still won't help with the range version
<K-ballo> I no longer understand what the concern is, what I've been imagining so far apparently isn't, and I can see you are starting to get frustrated
<K-ballo> you should probably wait for hkaiser to pop up, and ask advice on how to proceed
<K-ballo> I don't see anything immediately wrong with a lambda.. but I'm just going with your std::bind snippet from above
ste||ar-github has joined #ste||ar
<ste||ar-github> [hpx] hkaiser pushed 1 new commit to khuck-patch-2: https://github.com/STEllAR-GROUP/hpx/commit/57ef5dbf1885531c3b5703d6f0a5c9c69b3840d0
<ste||ar-github> hpx/khuck-patch-2 57ef5db Hartmut Kaiser: Merge branch 'master' into khuck-patch-2
ste||ar-github has left #ste||ar [#ste||ar]
eschnett has joined #ste||ar
<brjsp> goodnight
brjsp has quit [Quit: Page closed]
hkaiser has joined #ste||ar
jaafar has quit [Ping timeout: 246 seconds]