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 quit [Quit: eschnett_]
parsa is now known as parsa_
Abhishek09 has quit [Quit: Page closed]
K-ballo has quit [Quit: K-ballo]
eschnett_ has joined #ste||ar
adityaRakhecha_ has quit [Ping timeout: 256 seconds]
hkaiser has quit [Quit: bye]
eschnett_ has quit [Quit: eschnett_]
Abhishek09 has joined #ste||ar
<Abhishek09> Hello guys l facing problems on installing blazr
<Abhishek09> Blaze
<Abhishek09> It saying cannot find HPXconfig.cmake
Abhishek09 has quit [Ping timeout: 256 seconds]
Abhishek09 has joined #ste||ar
<Abhishek09> hi guys i facing problem
<Abhishek09> in installing blaze
<Abhishek09> By not providing "FindHPX.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "HPX", but CMake did not find one.
Abhishek09 has quit [Ping timeout: 256 seconds]
Abhishek09 has joined #ste||ar
<simbergm> Abhishek09: I've never built blaze but you'll need to point cmake to your HPX installation
<Abhishek09> No i solved the problem
<simbergm> -DHPX_DIR=/path/to/hpx/installation/lib/cmake/HPX or something like that
<simbergm> nice
<Abhishek09> Now i am building phylanx on mojave
<simbergm> good luck!
mdiers_ has quit [Remote host closed the connection]
<Abhishek09> i have successfully installed
<Abhishek09> even on mojave
Abhishek09 has quit [Quit: Page closed]
david_pfander has joined #ste||ar
<simbergm> `make test` might be a good idea...
<simbergm> or whatever they have set up to run tests
david_pfander has quit [Ping timeout: 246 seconds]
mdiers_ has joined #ste||ar
rohit64 has joined #ste||ar
K-ballo has joined #ste||ar
david_pfander has joined #ste||ar
<zao> heller_: do we run any TravisCI? They just sacked a ton of their competent people after being acquired.
<zao> I wonder if Abhi there used a prefix or installed system wide, considering that I could not install Phylanx due to package building bailing out about prefixes...
Abhishek09 has joined #ste||ar
<Abhishek09> hey diehlpk_work: are you here?
<zao> Abhishek09: did you install phylanx into any particular prefix and where did you get the python from? My install failed and I would like to know why :)
<Abhishek09> i succesfully build it,
<Abhishek09> At begining have used homebrew
<Abhishek09> i faced problem at installing blaze but i resolve this problem
<Abhishek09> hey deihlpk_work: are you here:
Abhishek09 has quit [Ping timeout: 256 seconds]
rohit64 has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]
nikunj has joined #ste||ar
nikunj has quit [Client Quit]
eschnett_ has joined #ste||ar
hkaiser has joined #ste||ar
Abhishek09 has joined #ste||ar
eschnett_ has quit [Quit: eschnett_]
adityaRakhecha has joined #ste||ar
<adityaRakhecha> Abhishek09: how did you solve blaze problem?
<Abhishek09> have facing the same problem?
<Abhishek09> *you
<adityaRakhecha> Abhishek09: The error you mentioned which you got while installing blaze, somewhat same I am getting while building phylanx.
<Abhishek09> by default it set to wrong path.So we have to correct it to get rid of errors.
<Abhishek09> what errors you got show here
<K-ballo> what happened here https://www.openhub.net/p/hpx ?
<K-ballo> oh, we are stellar-hpx
<heller_> zao: no travis here
<zao> EasyBuild is heavily relying on Travis and might need to find something else.
<zao> How bad is Circle to set up? :)
<adityaRakhecha> Abhishek09: Did you get any errors during building phylanx? I mean after pybind11.
<heller_> zao: not bad at all
<Abhishek09> NO,you mean to say that you facing issue during pybind11
<heller_> zao: about the same
<adityaRakhecha> Abhishek09: No during phylanx.
<Abhishek09> Please paste the error here
<zao> In a gist
<zao> Not directly in chat :)
<adityaRakhecha> Abhishek09: Please check here -> https://gist.github.com/adityaRakhecha/28ff87a9b2646f349c6d7af1e05a4f50
<adityaRakhecha> zao: :)
<Abhishek09> hey aditya:search file pybind11Config.cmake in your pc
<Abhishek09> did you find aditya?
<adityaRakhecha> Abhishek09: yes.
<Abhishek09> Where it is?
<Abhishek09> which os you are using?
<adityaRakhecha> Ubuntu 16.04
<Abhishek09> Aditya:^^^
<heller_> You should upgrade...
<heller_> Anyway, the error tells you what to do. Same procedure as with hpx
<Abhishek09> aditya:what happens r u solved that bugs?
<heller_> Which bug?
<Abhishek09> he is facing with issues?
<hkaiser> Abhishek09: we discussed this yesterday at length
<hkaiser> the issue is on your side, not ours
<Abhishek09> yes but i already solved
aserio has joined #ste||ar
<heller_> So we're not talking about a bug, but a problem on your side?
<heller_> If you solved it, great!
eschnett_ has joined #ste||ar
<heller_> What's the problem then?
bibek has joined #ste||ar
<diehlpk_work> Abhishek09, Yes, I am back
<Abhishek09> diehlpk_works: i have installed phylanx and hpx on mojave.
* Yorlik curses@curses
<Yorlik> Merging curses and streams ... U&$&%$§%
<zao> Yorlik: As in ncurses? Oh boy.
<zao> (at least you're not on AIX or Solaris, theirs are worse)
<Yorlik> Actualy PDCurses
<Yorlik> I want some very BASIC console IO for my server, lik,e a status bar, a scroll area, an input area
<Yorlik> Actually I already married cout to a curses window already
<Yorlik> But the scrolling is a little pain - I never used curses interface before
<Yorlik> It's not bad .. just .. all this boilerplate coding .. lol
<Yorlik> I thinbk on the long run it will help
<Yorlik> It just seems O have to do all my line overflows and row overflows and stuff m,anually ..
<Abhishek09> hey diehlpk_works:can i create bunch of .whi files for all dependencies.
<Abhishek09> *whl
aserio has quit [Ping timeout: 264 seconds]
<heller_> hkaiser: what about #3662?
aserio has joined #ste||ar
<hkaiser> heller_: will have a look
<heller_> thanks
<heller_> btw, openmp5 now has events
<heller_> almost futures ;)
akheir_ has quit [Quit: Konversation terminated!]
<parsa_> Abhishek09: no. you're supposed to create only one wheel file that installs phylanx and contains all of its non-python dependencies
parsa_ is now known as parsa
akheir has joined #ste||ar
<diehlpk_work> Abhishek09, Cool, does the test run?
<hkaiser> heller_: I know, let's see if anybody implements those ;-)
<hkaiser> heller_: now tell me again that we don't have any impact
<zao> How hard could it be? :P
<hkaiser> zao: what implementing things?
<zao> (kneejerk reaction to OpenMP mentions :D)
<Abhishek09> Yes@deihlpk
<heller_> hkaiser: i talked to the new CEO ... the influences are: OMPSs and StarPU, according to him
<Abhishek09> i will be back in 30 min
<hkaiser> sure
<zao> So how's master nowadays?
<hkaiser> heller_: and where do you think those guys have those ideas from ?
<heller_> quote: "There is no industry standard offering the same functionality and power, only openmp5"
<zao> heller_: Our NLAFET Horizon 2020 project has gone for StarPU... not sure if they're happy about it or not :)
<heller_> ompss had their task dependencies before we had them, IIRC
<hkaiser> +
<heller_> zao: tell them they should make a study and compare it with HPX ;)
<diehlpk_work> Abhishek09, Ok, what we want is to know if it is possible to have one wheel for phylanx and do pip install phylanx
<zao> I like their PhDs too much :)
<heller_> :P
<hkaiser> zao: they want to go with something European, the US is very much hated anyways
<zao> I see that you're well versed in project politics :)
<heller_> since when is nuremberg, lugano or zurich in the US now ;)
<zao> Gonna be fun to see how it goes for them with brexit, considering that half the project is UK-based.
<hkaiser> heller_: you know that the US from inside is different from what it looks from the outside
<heller_> lol
<hkaiser> zao: nothing will change there, no worries
<heller_> starpu is INRIA, isn't it?
Abhishek09 has quit [Ping timeout: 256 seconds]
<heller_> I am very happy to not having to do H2020 projects in the near future again
<hkaiser> why, it's good money
<heller_> is it worth it?
<zao> Is it worse than PRACE?
<heller_> never did PRACE
<hkaiser> heller_: well, for you currently not
<heller_> right
<heller_> that's what I mean
<diehlpk_work> parsa, https://github.com/pfultz2/cget This could be nice for installing the C++ dependencies
<hkaiser> diehlpk_work: not sure about this, there are other options
<heller_> yes
<hkaiser> vcpkg, conan
<heller_> and we probably should not intertwine our build system with dependency management
<hkaiser> right
<heller_> we don't want to maintain that
<heller_> It's not _that_ hard
<hkaiser> heller_: well, if it was easy we would have a generally used solution
<parsa> diehlpk_work: using spack or vcpkg would be easier if we want to go that route
<diehlpk_work> Why would cget affect our build system?
<heller_> hkaiser: it is hard to find a general solution. It is not that hard to learn something that works for you
<hkaiser> nod
<heller_> spack, vcpkg, all those options are there and exist
<parsa> because we have to maintain it to work with change in hpx or others
<parsa> changes*
<diehlpk_work> Let us wait what the GSoC student present for a pip package
<parsa> yeah
<hkaiser> heller_: I have some changes for #3662 (Windows specific) - ok to push to your branch?
<heller_> hkaiser: sure
<heller_> hkaiser: still got segfaults there?
<heller_> I need a ghost writer
<hkaiser> heller_: no, removing the #include to the header you deleted, etc.
<heller_> ahh, sorry about that
<nimishkhurana[m]> I think canon can be used for making package for phylanx?
<heller_> got a positive on the windows builder on packet, btw
<heller_> *conan
<nimishkhurana[m]> *conan
<hkaiser> heller_: marvelous!
<heller_> nimishkhurana[m]: why not just use the infrastructre pip provides?
<heller_> hkaiser: I just need to pass this one last exam next week ;)
<hkaiser> yah, sure
<hkaiser> shouldn't be a problem, just an hour of babbling
<heller_> 1.5 but yeah
<hkaiser> heller_: how did your talk go, btw?
<heller_> slides need to look nice
<heller_> it was nice
<heller_> the live demo was successful and had the intended effect
<hkaiser> cool
<heller_> 2x on skylake, 5x on naples
<hkaiser> what did you show in the demo?
<heller_> very nice
<heller_> numa placement with HPX
<heller_> the standard stuff
<hkaiser> could you show me the code for that?
<heller_> sure
<heller_> it's this saxpy code
<heller_> one sec
<heller_> the packet guys will ask us for a feature on their site soon, I think
<hkaiser> heller_: #3662 is good to go from my end
<heller_> yay
<hkaiser> heller_: nice and small example
<heller_> yup
<hkaiser> mind if I use that?
<heller_> please do
<hkaiser> heller_: do you have to use hpx::compute::vector? wouldn't a std::vector<double, allocator_type> do as well?
<heller_> IIRC, no
<hkaiser> why not?
<heller_> because of the first touch thingy we do
<heller_> but I am not sure if we do that in allocate or construct..
<hkaiser> I need to verify this
<heller_> we really have to switch to hwloc2...
<heller_> yeah
<hkaiser> what does hwloc2 give?
<hkaiser> late placement?
<heller_> direct placement, yes
<heller_> oh btw, I also have some new information/clarification on the new dataflow execution engines that we'll see popping up by the end of the year
<heller_> unfortunately, they won't be as awesome as we were hoping them
<heller_> so they wont't support dynamic task execution, only static dataflow graphs
<hkaiser> gah
<hkaiser> compiler generated graphs, I presume
<heller_> yup
<heller_> and a big problem if the code is larger than the nodes your hardware has
<heller_> my guess is that A21 will be a LINPACK machine so that some people don't loose their faces
aserio has quit [Ping timeout: 264 seconds]
hkaiser has quit [Read error: Connection reset by peer]
hkaiser has joined #ste||ar
<hkaiser> heller_: absolutely
Abhishek09 has joined #ste||ar
<Abhishek09> parsa:i think cibuildwheel is suitable for my project.
<zao> Hrm
<heller_> is there a document on how to build pip packages?
<heller_> and what to do with external C/C++ libraries?
<Abhishek09> Yes,there is documentation
<zao> Seems like custom is to bundle everything into the final wheel itself, or attempt to static link
<zao> Abhishek09: that still just does manylinux1 and macOS
<Abhishek09> but i can't use manylinux1
aserio has joined #ste||ar
aserio1 has joined #ste||ar
<zao> Exactly :)
<aserio1> hkaiser: would you be interested in joining the tiling meeting?
<nimishkhurana[m]> Can't it be done by simple setuptools https://setuptools.readthedocs.io/en/latest/setuptools.html#id15 and giving urls of non-PyPI dependencies like boost,hwloc,etc or making wheels for them individually .
aserio has quit [Ping timeout: 264 seconds]
aserio1 is now known as aserio
<diehlpk_work> nimishkhurana[m], giving urls would not help to build and install them
<nimishkhurana[m]> It is from that documentation only. What does that means ?
<diehlpk_work> nimishkhurana[m], It would be your task ti understand that
<Abhishek09> diehlpk_work:we have to modify them after creating .whl
<diehlpk_work> I do not know, if I would know how to do generate the pip package, it would not be GSoC project
<diehlpk_work> nimishkhurana[m], Abhishek09 Why not start with a package, which have hwloc as one of the dependencies
<diehlpk_work> hwloc is easy to compile and you could test that
<nimishkhurana[m]> diehlpk_work: I understood that..it says you can provide that download url of non pypi dependencies or the url of webpages which contain download url.
<Abhishek09> diehlpk_work: but you said earlier that i am moving in right direction
<nimishkhurana[m]> Will try that.
<diehlpk_work> Abhishek09, What do you mean?
<Abhishek09> i have said earlier that we have create binary wheels then we bundle them into package.
<Abhishek09> we do not have to worry about setup.py
<diehlpk_work> Abhishek09, There are probably different solutions to do this
<diehlpk_work> As I told you before in the end you and all other students have to submit their own proposal how to generate a pip package
hkaiser has quit [Read error: Connection reset by peer]
hkaiser has joined #ste||ar
<hkaiser> aserio: it's probably too late now...
<aserio> hkaiser: I don't know... we are still going strong :p
<hkaiser> do you need me there?
<aserio> hkaiser: no, I don't think so
<hkaiser> good, good
Abhishek09 has quit [Ping timeout: 256 seconds]
Abhishek09 has joined #ste||ar
aserio has quit [Ping timeout: 258 seconds]
aserio has joined #ste||ar
<Abhishek09> how to extract lib files from required package
iti_ has joined #ste||ar
<diehlpk_work> Abhishek09, Be more precise?
bita has joined #ste||ar
<hkaiser> aserio: yt?
<aserio> hkaiser: yes
<hkaiser> Could you tell Maxweel to call me (skype), pls?
<iti_> I was reading task description for Domain decomposition and load balancing for crack and fracture mechanics code, and it's written there that -> " A code utilizing the parallel algorithms and futurization within HPX for a single shared memory node is available." Can't seem to find the code, any lead? @diehlpk_work?
<Abhishek09> hey diehlpk_works: can i use yum?
<aserio> hkaiser: done!
<hkaiser> thanks!
<diehlpk_work> Abhishek09, yum is very outdated, but dnf could be fine
<Yorlik> Does anyone of you have experience with merging streams into C style IO? I am using PDCurses as a cross platform curses implementation and managed to make a custom stream with a custom streambuffer whgich can take cout and a curses wirndo and output the cout stream to the window. But there is some oddities and problems. One of them is my cin input not working anym ore at all (guess I'll change it to either getch()
<Yorlik> or find a custom stream recipe too). Since HPX uses C++ streams for output and the overall stream concept has several advantages I'd rather not do all of my console IO in C style and rather find a way make specialized streams work with pdcurses.
<Yorlik> Ideas ?
<hkaiser> should be easy to implement using boost iostreams
<diehlpk_work> Abhishek09, This would only work for fedora or
<Yorlik> hkaiser: What do boost iostreams offer over the std iostreams?
<hkaiser> aserio: could you ask Maxwell to go to appear.in/stellar-group ?
<hkaiser> Yorlik: you can implement custom streams easily
<aserio> one sec
<diehlpk_work> iti_, The code is not available yet, but we would provide you with a small toy code
<aserio> hkaiser: delivered
<hkaiser> thanks
<Abhishek09> deihlpk_work:i think it doesnt suport for mojave
<iti_> I see, so before writing proposed solution it would help to have a code (even toy code) to understand better approaches. Would that be possible?
<iti_> I mean would it be possible for you guys to provide that during application period?
<diehlpk_work> Abhishek09, Yes, but using dnf could probably not work, because I do not think that phylanx will compile with hpx 1.2.1
<diehlpk_work> iti_, Sure, but you could look into this paper here https://www.sciencedirect.com/science/article/pii/S0010465508002348
<diehlpk_work> Probably, the easiest thing would be to implement the code by yourself in C++
<diehlpk_work> So you would have basic understanding of the code
<iti_> Sounds good, thanks for the reference paper! Should definitely begin by collecting information about it. :D
<diehlpk_work> yes, the code itself is not difficult it is just a bunch of for loops
<Abhishek09> why phylanx will not compile with hpx 1.2.1.
<diehlpk_work> The task would be to distribute this code to many nodes
<diehlpk_work> And later work on load balancing, if you have enough time
<iti_> Yes, right. I'll implement it then, I did read about the task description. Looks interesting to work on, let's see what I get to in coming week :)
<diehlpk_work> Abhishek09, I assume that phylanx needs newer features as in hpx 1.2.1
<iti_> Thank you again, will be back after reading the paper and more info on the same! Have a good day.
<diehlpk_work> iti_, Sure, computing the neighbors, updating the forces, and positions is not difficult to implement and you would understand how the code works.
<diehlpk_work> iti_, Sounds good. Bye
iti_ has quit [Quit: Leaving]
<Abhishek09> according to you then i have to use yum!
<diehlpk_work> On which operating system?
<Abhishek09> tell if you have problem then i can switch to linux
<diehlpk_work> You use yum on Mac OS?
<Abhishek09> otherwise i work on mojave
<Abhishek09> yes
<bita> hkaiser, using phylanx.execution_tree.var() I get a new error running pytest on phylanx backend. Do you have the time to look at it?
<hkaiser> bita: sure
<bita> this is the error "NotImplementedError: Phylanx does not support non-NumPy member functions.Cannot transform: phylanx.execution_tree.var" calling `phylanx.execution_tree.var(np.eye(size, dtype=dtype))`
<diehlpk_work> Abhishek09, Are we speaking about this yum https://en.wikipedia.org/wiki/Yum_(software)
<diehlpk_work> If so I have no clue how you get the rpm packages compiled for Mac OS?
<Abhishek09> No it is package management tool.
<Abhishek09> it install a system package required by our library.
<diehlpk_work> Abhishek09, I doubt you can install hpx with whatever yum you are talking about
<Abhishek09> No,we have extract library files needed by hpx.
<hkaiser> bita: would you mind creating a ticket for this?
<hkaiser> I'll have a look asap
<bita> of course, thanks
<diehlpk_work> Abhishek09, Yes, but how you would compile hpx?
<hkaiser> bibek: I think the code above shouldn't be in a @Phylanx decorator
<hkaiser> you should execute the phylanx.execution_tree.va(np.eye(...)) outside of a @Phylanx decorator
<hkaiser> bita: ^^
<diehlpk_work> Abhishek09, I would suggest you come up with a solution and I will let you know if it is agood one
<diehlpk_work> Please prepare a proposal and I am happy to look into it
<hkaiser> bita we need to think how to capture the phylanx execution tree such that the Keras frontend can access it
<Abhishek09> By Packing all external shared libraries into the wheels
<bita> I see
<hkaiser> bibek: also, if you want to execute the Phylanx eye() function, you don't need to wrap it in a var()
<hkaiser> the decorator 'wraps' the results of the invocation into a var() already
<bita> That's right, thanks, Now I get a different error "NotImplementedError: Only the followig are acceptable Phylanx array types:
<bita> E booleans: ['b', 'bool']
<bita> E floats: ['f', 'f2', 'f4', 'f8', 'float']
<bita> E integers ['u2', 'u4', 'u8', 'i', 'i2', 'i4', 'i8', 'int']"
<Abhishek09> diehlpk_work:my approach is correct or not?
<diehlpk_work> Abhishek09, Please prepare the proposal and I can have a look
<Abhishek09> but where i can share as application period doe not starts
<diehlpk_work> As far as I know you provide the link to a google document or upload a pdf file
<diehlpk_work> Share a google document with me and write down the basic idea
<diehlpk_work> Like how things are done. Where do you get the libs to include in the package from
<diehlpk_work> How things are compiled
<diehlpk_work> For which OS will it work
<diehlpk_work> and so on
<Abhishek09> ok how many slots for GSoC is available this year?
<diehlpk_work> Abhishek09, have you read the rules of GSoC and the time line?
<Abhishek09> Yes,ofcourse
<Abhishek09> but i have confusion about slots availability
<diehlpk_work> Ok, your question does not make sense, since we even not know that we got accepted
<Abhishek09> i think previous year ste||ar given 2 slots.
<zao> Must be because we’re so awesome
<diehlpk_work> Abhishek09, I see 7 last year
<Abhishek09> organisation will anounced on 27th feb
<diehlpk_work> Abhishek09, February 26 12:00 UTC List of accepted mentoring organizations published
<Abhishek09> vary with timeline
<hkaiser> bitatalk to Rod, it's his error
<hkaiser> bita ^^
Malvolio has joined #ste||ar
<bita> hkaiser, I will
aserio has quit [Ping timeout: 264 seconds]
bibek has quit [Quit: Konversation terminated!]
akheir has quit [Quit: Konversation terminated!]
eschnett_ has quit [Quit: eschnett_]
aserio has joined #ste||ar
<Abhishek09> diehlpk_work:is it necessary to install hpx or i can use lib files
<aserio> hkaiser: yt?
akheir has joined #ste||ar
<hkaiser> aserio: here
<aserio> hkaiser: two questions, one: Vinay has proposed May 7th for a visit to PNNL, does that work for you?
<hkaiser> I think so, it's a bit earlier than anticipated, but should work
<hkaiser> I'll respond
<aserio> I would not be opposed to pushing it back a week
<hkaiser> nod
<aserio> two: I created a version of the lw_1d which uses replay, should I push this to hpxr repo
<hkaiser> too late :/
<aserio> lol
<hkaiser> aserio: wow!
<hkaiser> sure - add it as an example
<aserio> hkaiser: I was wondering about licensing though
<hkaiser> aserio: is the 7th a nonono or just a no from from your end
<hkaiser> I can still backtrack, call myself an idiot etc
<aserio> Cory's Birthday is on the 8th, but I think we will be back that evening anyway
<hkaiser> should be, yes
<aserio> I could send an email asking anyway
<hkaiser> ok, fine
<aserio> *asking to move it
<aserio> hkaiser: I will push my changes to a branch with the labs licensing
<aserio> that away if we need to change things I can delete it
<aserio> the repo is also private atm
<hkaiser> right, Keita said is has to stay private for the time being
<hkaiser> aserio: was it difficult to change things?
<aserio> hkaiser: not really, but it is a hacky change
<hkaiser> well, it's a first step
<aserio> I want to go back and change the structure of the code a bit to use our semantics
<aserio> otherwise the dataflow_replay is a cool feature
<hkaiser> nod, fits nicely into our semantics
aserio has quit [Quit: aserio]
Abhishek09 has quit [Ping timeout: 256 seconds]
Abhishek09 has joined #ste||ar
<Abhishek09> hi guys
<Abhishek09> is anyone here
<Yorlik> WOW: string x = 0; --> exception, no compiler error ..
<Abhishek09> i have seen two blaze
<hkaiser> Yorlik: yah, you should get a warning, though
<Yorlik> I had switched the type from char to string but not adjusted the assignment
<Abhishek09> one on github and another one in bitbucket
<hkaiser> Yorlik: it probably uses the constructor that is meant to copy from a given C-style string: string s = "some blurb";
<Abhishek09> hkaiser:is both are same?
<hkaiser> Abhishek09: the bitbucket one is the correct one
<Yorlik> Yes - makes sense - it just didn't expect 0 :)
<hkaiser> well, if yougive it a nullptr you get back what you deserve ;-)
<Abhishek09> but is non pypi.
<hkaiser> Abhishek09: what do you mean by 'but is non pypi'?
<Abhishek09> github one is pypi
<Abhishek09> hkaiser:can i use github one to compile phylanx?
<hkaiser> shrug, never tried
<hkaiser> I don't know what pypi is
<hkaiser> is that a Python thing?
<Abhishek09> pypi means instalable through pip
<hkaiser> why should Blaze be installable through pip? It's a C++ library after all...
Abhishek09 has quit [Quit: Page closed]
parsa is now known as parsa_
eschnett_ has joined #ste||ar
parsa_ is now known as parsa
parsa is now known as parsa_
parsa_ is now known as parsa
<Yorlik> I wish there were a crossplatform c++ equivalent to curses ..
<Yorlik> Trying to marry curses with streams is just horrible.
* Yorlik curses
jbjnr__ has joined #ste||ar